N. A. Kashkarov,
Professor of the Tomsk Technological Institute of Emperor Nicholas II.

Modern

Methods of Water Purification.

With an appendix:

Biology of Drinking and Waste Waters.

Price 3 rubles.

Tomsk.
Typo-lithography of the Siberian Association of Printing, corner of Dvoryanskaya St. and Yamskoy Lane, own building.
1912.

CONTENTS

	Page
Preface	1
Chapter I.	3

Properties of water: physical (3), chemical (8), biological (micrographic (19) and bacteriological (24)).

Influence of drinking water on the spread of epidemics and on the mortality of the population (30). Requirements imposed on drinking water (36).

	Page
Chapter II.	39

Classification of methods of water purification (39). Removal of suspended substances.

Straining of water (41). Sedimentation (42).

	Page
Chapter III.	55

Filtration of water (55).

English sand filters; general description of the arrangement (56). Theory of the action of sand filters (56); role of sand (59); mechanical theory of the action of filters (60). Change in the chemical composition of water (63). Influence of the thickness of the filtering layer, the grain size of the sand, and the rate of filtration on the operation of the filter (64). Construction and calculation of English filters (66). Loss of head during filtration (74). Devices for admitting and discharging the filtered water (75). Appurtenances of the filter (79). Open and closed filters (82).

Operation of English filters (83). Instructions for the management of filters in Berlin (83). Cleaning of filters (88). Washing of sand (93). Cost of installation and operation (96). Merits and shortcomings of English filters (96).

	Page
Chapter IV.	98

Modifications of the English method of filtration.

Filtration with prefiltration (98).

Puech-Chabal filters (103).

Cases of application of double filtration and Puech filters (108).

Intermittent filtration (108).

Non-submerged filters (109).

II

Chapter V. ........................................................................ 110

American (mechanical) filters.

Coagulation (111). General description of the arrangement of American filters (113). Various systems of American filters: Jewell (118), Howatson (120), Rives (120), New York Company (120), Bell (121). Kreincke filters (122) and Zimin filters (122).
Large filtration stations of the American system (Cincinnati) (123).
Results of water purification by American filters (127).
Cost of construction and operation of American filters (128).

Chapter VI. ....................................................................... 131

Filters of special systems.

Fischer and Peters filters (131), Kurka (133).
Chamberland domestic filters (133) and Berkefeld (135).
Howatson filters (135), Candy filters (135).
Anderson system (137).

Chapter VII. ...................................................................... 139

Change in the content of certain dissolved substances in water.

Aeration of water (139).
Softening of water (140). Principles of softening methods (140). Chemical treatment of hard water (141); determination of the quantity of reagents (142). Beranger and Stingl, Derveaux, Dermo apparatus (145). Softening of water by means of zeolites (148).
Reduction of the acidity of water (151).
Removal of iron from water (deferrization) (152).
Removal of manganese from water (160).
Irrigation of water (161).

Chapter VIII. ..................................................................... 164

Physical methods of disinfection (sterilization) of water.

Disinfection by ultraviolet rays (165). Characteristics and properties of ultraviolet rays (165). Apparatus for disinfecting water (168).

Chapter IX. ....................................................................... 172

Chemical methods of water disinfection.

Disinfection of water with hypochlorites (173); nature of the process (174); required quantity of chlorine (175) and duration of the action of chlorine on water (175); place where the reagent is added (176).
Results of treating water with chlorine (180).
Cost of treating water with chlorine (182).

III

Advantages and disadvantages of the method (183).
Use of sodium hypochlorite (185).
“De-clor” system (185).
Duck Howatson system (186).

Chapter X. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

Ozonation of water.

Ozone and its production (187). Siemens ozonizers (190), Otto (193), Gérard (196), de Frise (196). Vasmer (197). Determination of the amount of ozone (198). Comparison of various ozonizer systems (199).

Ozonation of drinking water (200). Sterilizers of Otto (200), de Frise (202), Siemens (202), etc. Comparison of various sterilizer systems (204).

Results of treating water with ozone (208). Influence of the properties of water on the results of ozonation (208). Ozone concentration and contact time (211). Control of results (212).

Ozone stations in St. Petersburg (212) and Hermannstadt (218).
Apparatus for domestic ozonation of water (219).

APPENDIX.

Biology of drinking and wastewater . . . . . . . . . . . . . . . . . . . . 221

Biology of drinking water . . . . . . . . . . . . . . . . . . . . . . . . . 225

Spring water (225). Groundwater (226). Water from shallow wells (231). Water from reservoirs (231). Water from lakes (233). River water (235).

Biology of wastewater and open bodies of water . . . . . . . . . . . . . . 237

Municipal wastewater, irrigation fields, artificial biological purification (237). Wastewater of factories and plants (242). Self-purification of bodies of water, especially rivers (244); polysaprobic zone (244), mesosaprobic (247), and oligosaprobic (251). Streams and ditches (254). Ponds (255). Shores of seas and lakes (257).

Methods of biological investigation. Organisms . . . . . . . . . . . . . . 258

Division of a body of water into regions (258). Apparatus for taking samples and investigation (260). Organisms: polysaprobic (264), mesosaprobic (258), and oligosaprobic (272).

Preface

Questions of water supply in general, and of the purification of drinking water in particular, are at the present time acquiring enormous practical importance, especially for Russian cities. According to data collected in 1911 by the “Permanent Bureau of Russian Water-Supply Congresses,” of the 1,082 cities of Russia only 192, i.e. about 17.7%, have water-supply systems, whereas, for example, in Germany water-supply systems have been installed in more than 70% of the total number of cities. Yet even among these 17% many cities—if not the majority—use water of very doubtful quality. A vivid example was provided by the cholera epidemic of 1908–9 in Petersburg, which was entirely caused by the unsatisfactory purification of Neva water. This epidemic apparently aroused interest in questions of water purification not only among specialists, but also in society at large.

Meanwhile, in the Russian language there is no systematic exposition of modern methods of water purification; and even in the foreign literature of recent times there are no systematic works on this question, although precisely in recent times the theory and practice of water purification have been undergoing particularly vigorous development. With the expansion of the requirements imposed on drinking water by hygienists, new methods of water purification arise, as well as a new form of purification—disinfection of water, whose aim is the destruction of pathogenic bacteria in water; following the spread of water ozonation, which is gaining predominance in Europe, the method of purifying water with hypochlorites, now adopted in almost 200 cities, has developed rapidly in America since the end of 1908; finally, ultraviolet rays are being studied, promising a cheap and elegant method of rendering water harmless. Alongside this, the role of the old methods of water purification, and with it their application, is changing. The development and clarification of the theoretical requirements imposed on drinking water and on the various processes of its treatment belong to hygienists, physicians, and natural scientists; they work out the general tasks, the technical implementation of which is undertaken by engineers. A vast literature is being created, scattered through various journals, both technical and others (for example, those devoted to questions of bacteriology).

The present work is an attempt to systematize modern methods of water purification; it is intended for technicians in the water-supply field, and therefore it sets forth certain information, wh—

...for the naturalist’s laboratory might have seemed excessively elementary and superfluous. The properties of water and the requirements imposed on drinking water at the present time are considered first.

The materials used in compiling the present work are indicated for the most part in the notes to the text; listed separately are only the major works on water supply. The sources are not indicated for those data that we collected during a trip abroad in the summer of 1911.

List of the principal sources used by the author:

Prof. Timonov. Water Supply and Watercourses.
Pr. Provodnik. Course in Water Supply.
G. Fleger. Wasserversorgung der Städte.
Frühling und Oesten. Wasserversorgung der Städte.
Débauve et Imbeaux. Assainissement des villes. Distribution d’eau.
Bechmann. Assainissement des villes.
J. Don. The filtration and purification of potable supply.
V. L. Omeliansky. Fundamentals of Microbiology.
Rubner, Gruber und Ficker. Handbuch der Hygiene.
Proceedings of the I–IX Russian Water-Supply Congresses.
Journals: Gesundheits-Ingenieur, La Technique Sanitaire, Sanitary Record, Engineering News, Engineering Record, Engineering, Municipal Journal and Engineer, Zeitschrift für Gasbeleuchtung und Wasserversorgung, Zeitschrift für Hygiene, Zentralblatt für Bakteriologie, Wasser und Abwasser, Génie Civil, and others.

Descriptions of the water supplies of various cities (municipal publications).

Works on individual questions, indicated in the notes to the text.

Chapter I

Properties of Water

Chemically pure water (H₂O) is a body consisting of a compound of 2 atoms of hydrogen and 1 atom of oxygen, or, by weight, of 2 parts hydrogen to 16 parts oxygen; thus, in a given quantity of water by weight there is 88.88% oxygen and 11.12% hydrogen.

By volume, 2 volumes of hydrogen, combining with 1 volume of oxygen, form 2 volumes of water vapor; at a temperature of 0° and a pressure of 760 mm, 1 liter of oxygen (O) weighs 1.4298 g, while 1 liter of hydrogen (H) weighs 0.08936 g; a liter of water vapor weighs 0.805 g.

Before examining the properties of water, let us point out that water, as a source of water supply, is divided by origin into 3 types: atmospheric waters (rain, snow), surface waters (water of rivers, lakes, seas), and underground waters (groundwaters, obtained from various kinds of wells, and spring waters, extracted directly at the place where the water emerges onto the surface of the earth).

In all these forms, water in nature is never chemically pure. Water usually contains various kinds of impurities, floating in it in an undissolved state (suspended impurities), or else dissolved impurities, which determine its chemical composition and external physical properties, as well as lower animals and plants, detectable with a microscope, and bacteria, whose presence can be determined by bacteriological examination.

Physical Properties of Water

By the physical properties of water are usually meant the so-called external characteristics: temperature, taste, odor, color, turbidity, and transparency.

Temperature.

Water in nature occurs at various temperatures, from temperatures close to 0° up to boiling water in some hot springs. But for use in the household and for other needs, only water of moderate temperature is suitable; thus, drinking water should, as far as possible, have a temperature from 7 to 12°, so as not to seem cold in winter and warm in summer.

Water colder than 5° is unpleasant and even harmful, since cold can cause irritation of the oral cavity, larynx, and digestive organs, as well as disorders of the liver. Water warmer than 14–15° ceases to refresh.

Taste and odor. Pure water is tasteless and has no odor; taste and odor arise from dissolved or suspended admixtures in the water. Most mineral salts, sugar, etc., impart taste to water without producing odor, whereas dissolved gases and oils act more upon the sense of smell than upon taste, although we often attribute the impression from them to taste.

The solubility of gases in water varies and decreases with rising temperature; the exception is hydrogen, which has constant solubility at temperatures of 0°–20° C.

Table 1 gives the solubility coefficients of several gases at a pressure of 760 mm.

Table 1.

Name of gases	Solubility coefficients at a pressure of 760 mm, in liters: t=0°	Solubility coefficients at a pressure of 760 mm, in liters: t=10°	Solubility coefficients at a pressure of 760 mm, in liters: t=20°	Specific gravity at 0° (relative to air)	Weight of 1 meter at 0° and pressure 760 mm
Air	0,025	0,020	0,017	1	1,2932 gr.
Oxygen	0,041	0,032	0,028	1,1056	1,43
Hydrogen	0,019	0,019	0,019	0,06929	0,0895
Nitrogen	0,020	0,016	0,014	0,9714	1,256
Carbonic acid gas	1,80	1,18	0,90	1,529	1,977
Carbon monoxide	0,032	0,026	0,023	0,968	1,254
Marsh gas	0,054	0,044	0,035	0,558	0,716
Hydrogen sulfide	4,37	3,59	2,90	1,171	1,523
Ammonia	1,049	0,812	0,654	0,597	0,761
Chlorine	3,00	2,58	2,16	2,47	3,81

Let us recall here the laws of solubility of gases:

1) At the same temperature, solubility increases directly in proportion to the pressure;

2) when a mixture of gases is in contact with a liquid, each gas dissolves in it as though it alone occupied the entire volume of the gas mixture.

Odor is often caused by the presence in water of microorganisms, living or dead, chiefly algae (algae), which develop especially strongly in lakes, ponds, and also on sand filters. Some of the living microorganisms produce aromatic odors similar, for example, to the odor of geranium (Asterionella) or violet (Mallomonas); others produce the odor of grass (Anabaena, Melosira, etc.) or fish (Chlorophyceae and some

Protozoa). After death, microorganisms, decomposing, impart to the water the odor of hydrogen sulfide or hydrocarbons.

It has not yet been fully established whether water with such an odor, or even with the odoriferous microorganisms themselves, is harmful to health. Apparently, they are harmless for a healthy person, but may be harmful when the intestine is already disordered.

In practice, to determine the absence of odor, 0.5 liter of water is taken and heated in a flask to 40–50° C.

If the water contains hydrogen sulfide, which masks all other odorous gases, a solution of copper sulfate is added to the water; copper sulfate, with free hydrogen sulfide (liberated from solution upon heating), forms copper sulfide and frees the water from this gas; the heating is then repeated in order to determine the other odorous gases.

There are no scientific methods or expressions for characterizing the strength of an odor, and one has to define an odor as “weak,” “noticeable,” “strong,” and so forth—that is, according to subjective impressions. A. Gérardin proposed determining the odor caused in air by organic admixtures by means of potassium permanganate and establishing the concept of the “ozonometric degree.” The ozonometric degree is the quantity (in milligrams) of crystalline oxalic acid that produces on potassium permanganate the same effect as the organic substances contained in 1 gram of air.

Taste. Taste sensations are divided into four groups: sweet taste, salty, sour, and bitter. The human tongue is not very sensitive, and the majority of salts admixed with water are perceptible to the taste only when the admixture is not less than 0.5–1 g per 1 liter; an exception is formed by salts of iron and copper, which impart a metallic taste at an admixture of 0.05–0.06 g per liter. Organic substances usually act almost exclusively on the sense of smell, and not on taste.

It seems to us that water has a “pleasant” taste when it produces upon us a refreshing effect, which depends on the temperature of the water and on the quantity of air and carbonic acid dissolved in it.

Color, turbidity, and transparency. One should distinguish: color, i.e., the hue inherent in the water itself, independently of the substances suspended in it; turbidity (“turbidity”), i.e., a greater or lesser degree of contamination of it by small particles suspended in it; finally, the transparency of water, which is the result of the two preceding properties and is explained by the light-transmitting capacity of water.

Color. Chemically pure water is colorless at shallow depth, and has a blue hue, passing into greenish, at a depth of several meters.

Any other color is due to substances dissolved in the water. These substances are for the most part of vegetable origin and come from leaves, moss, and grass, which sometimes impart a yellowish coloration to surface waters.

The coloration of ground waters, which in general is rarely encountered, depends chiefly on soluble compounds of iron formed as a result of a lack of oxygen; on contact with air, these compounds, oxidizing, pass into insoluble forms and are precipitated from the water.

To determine the (relative) degree of coloration of water, it is compared with distilled water. For this purpose two identical cylindrical vessels of pure colorless glass are taken, 20–30 mm in diameter and 30–40 cm long, and one is filled with the water under examination, the other with distilled water; then one looks through the cylinders at a sheet of white paper placed beneath the bottom. Sometimes, instead of distilled water, a liquid of a definite coloration¹ is poured in to such a height that its column has the same transparency as a column of the water under examination of constant height. Instruments for determining color are called colorimeters².

The coloration of water is eliminated by removing the dissolved or suspended substances by which it has been caused.

Cloudiness. The water of most sources does not always contain the same quantity of suspended substances; in surface waters it usually increases in spring, during the melting of snows, when the runoff waters carry earthy particles away from the surface of the ground. However, not all suspended particles make water turbid: thus, bacteria, even in considerable numbers, do not cause turbidity, owing to their negligible size. In order to cause turbidity of water, the suspended particles must be sufficiently large and be present in sufficient quantity. Suspended substances are sometimes organic, sometimes inorganic (most often earthy). Organic substances enter almost exclusively into lakes and artificial reservoirs, as a result of the decomposition of algae and small animals; we shall return to them when considering the biological properties of water. Earthy particles may be siliceous and clayey; the former are easier to remove, while the latter are extremely fine and gelatinous (“colloidal clay”) and settle very slowly. The presence of colloidal clayey particles together with organic impurities explains the so-called opalescence (from the word opal) of river water during floods, i.e. the coloration of water resembling milky glass. The elimination of opalescence is very rarely possible—

can be achieved mechanically (by settling and filtration), it is more often necessary to resort to chemical purification of the water (so-called coagulation).

The degree of turbidity of water is characterized either by the quantity of suspended substances per unit volume of water, or by the transparency of the water; with the latter method, as indicated, the results obtained depend not only on the turbidity of the water, but also on its inherent coloration.

To determine the quantity of suspended substances, evaporation is used (for a description of the process, see below).

When determining transparency, instruments similar to the above-mentioned colorimeter are used, or else a column of the water under investigation, 70 cm high, is poured into a cylinder of transparent glass, and beneath the bottom a white plate is placed with figures of different sizes printed on it (Hiller’s scale); the smallest of the visible figures gives the degree of transparency of the water being investigated.

At the present time the degree of transparency of water is more often determined by the depth to which a platinum wire 1 mm in diameter, immersed in the water, is visible. At the suggestion of the American investigators A. Hazen and G. Whipple, as a standard for expressing the turbidity of water in degrees, water is taken that contains 100 mg of silica (amorphous), in such a state of subdivision that a platinum wire 1 mm in diameter is visible to a depth of 100 mm, with the observer’s eye situated 1.20 m above the wire; the turbidity of such a standard is designated as 100 degrees. To evaluate water more turbid than the standard, it is diluted with distilled water to the turbidity of the standard, and the ratio of the volume of diluted water to the original volume of turbid water is determined. To evaluate less turbid water, conversely, the standard is diluted.

By means of laboratory experiments, Hazen determined by the indicated method the turbidity of water in degrees corresponding to the depths to which a platinum wire of 1 mm is visible; this scale has been adopted in America.

Degrees of turbidity	Depth of platinum wire	Degrees of turbidity	Depth of platinum wire	Degrees of turbidity	Depth of platinum wire	Degrees of turbidity	Depth of platinum wire	Degrees of turbidity	Depth of platinum wire
	mm		mm		mm		mm		mm
7	1095	18	468	30	187	110	93	300	43.2
8	971	19	446	55	171	120	86	350	38.8
9	873	20	426	60	158	130	81	400	35.4
10	794	22	391	65	147	140	76	500	30.9
11	729	24	361	70	138	150	72	600	27.7
12	674	26	336	75	130	160	68.7	800	23.4
13	627	28	314	80	122	170	65.4	1000	20.9
14	587	30	296	85	116	180	62.4	1500	17.1
15	551	35	257	90	110	190	59.8	2000	14.8
16	520	40	228	95	105	200	57.4	3000	12.1
17	493	45	205	100	100	250	49.1	—	—

Influence of the physical qualities of water on health. Without doubt, it is desirable that drinking water be of a suitable temperature (7–12°C), without taste, without odor, and perfectly transparent. However, water that lacks one or another of these qualities is by no means always harmful to health; in each individual case it is necessary to determine the cause of a particular physical defect of the water, and only depending on this cause decide whether the water is harmful and whether, from the standpoint of hygiene, the defect that has been found must be eliminated. As an example, let us point out that water containing much iron becomes turbid upon contact with air, owing to the oxidation of the iron salts dissolved in it and their conversion into insoluble compounds, which precipitate from the water and make it turbid; nevertheless, such water is entirely harmless to health. Conversely, water that appears impeccably clean may contain many bacteria and be dangerous to health (let us point to the Neva water, which served as a vehicle of cholera in St. Petersburg, but was excellent in appearance and taste).

Nevertheless, despite all the assurances of hygienists, a person feels an instinctive distrust and even aversion toward water that does not possess physical purity—turbid, odorous, etc.; and if there is a possibility of choosing between water unpleasant in appearance, though known to be harmless, and water clean in appearance, even if containing bacteria, a person will almost always prefer the latter; in places where harmless water is unpleasant in appearance and there is no other water at all, there is observed an increased consumption of other beverages, for example alcoholic ones, which is unquestionably harmful.

Therefore, when supplying water, it is always necessary to take care to improve its external physical properties, to “beautify” the water (“Schönheit”), and not to confine oneself to making it harmless.

Chemical properties of water.

Chemically pure water (H₂O) is encountered in nature only very rarely.

The use of chemically pure water (for example, distilled water) is hardly beneficial for human beings. There is an opinion³) that the constant use of such water is directly harmful to the organism.

Although sailors do not suffer from using distilled water at sea⁴), nevertheless the dissolution in it of air (“aeration,” airing of water) or of carbonic acid, and the addition of a small quantity of certain salts, are unquestionably beneficial.

In natural water suitable for drinking, the following mineral substances are more often than others contained: 1) lime (CaO), 2) magnesia

³) Koeppe. “Deutsche Medizinische Wochenschrift,” 1898, 624.
⁴) Nocht. Hygienische Rundschau, 1892, p. 273.

(MgO), 3) gypsum (CaSO₄), 4) iron salts (Fe₂O₃), 5) potassium salts (KSO₄, KNO₃), 6) sodium salts (NaSO₄, NaNO₃), 7) nitric acid (HNO₃), 8) nitrous acid (HNO₂), 9) ammonia (NH₃), 10) sulfuric acid (H₂SO₄), 11) sulfurous acid (H₂SO₃), 12) chlorine (Cl), 13) silica (SiO₂).

In addition, there are organic substances and dissolved gases (especially carbon dioxide (CO₂) and oxygen (O), as well as hydrogen sulfide (H₂S)).

The solubility coefficients of gases are given above in Table No. 1, and of some salts—in Table No. 2.

Table II.

Name of compounds	Chemical designation	Solubility coefficient at pressure 760 mm	Temperature C°
Calcium sulfate	CaSO₄	0,0021	21
Potassium sulfate	K₂SO₄	0,103	15
Sodium sulfate	Na₂SO₄	0,163	18
Potassium chloride	KCl	0,326	15
Sodium chloride	NaCl	0,357	15
Magnesium chloride	MgCl₂	0,307	15
Potash	K₂CO₃	0,244	10
Soda	Na₂CO₃	0,083	10

Table III.

Maximum quantities of impurities in drinking water permitted by certain institutions and individuals.

Name of impurities	Name of units	Brussels Congress, 1885¹	Swiss Congress, 1888¹	Tiemann and Gärtner 1889¹	Klut 1909²
Total hardness	German degrees	20	—	18–20	—
Dense residue	milligr. in 1 liter	500	500	500	300
Sulfuric acid	”	60	—	80–100	60
Chlorine	”	8	20	20 30	30
Nitric acid	”	2	20	5–15	up to 30
Nitrous acid	”	—	0	0	0
Ammonia	”	—	0	0	0
Organic substances (determined with the aid of permanganic-acid potassium)	”	10	10	6–10	up to 12
Including organic carbon	”	—	—	5	—
Including organic nitrogen	”	0,1	0,05	0,2	—

Even those impurities which in small quantities are harmless to man may be harmful when they are present in large amounts in water.

¹ According to Frühling.

² Data from studies of the best drinking waters of Northern Germany.

Some investigators, such as, for example, Fischer, Reichardt, Tiemann, as well as the Brussels and Swiss hygiene congresses (1885), and many others, have attempted to assign approximately such maximum quantities of impurities in water, the exceeding of which makes the water unfit for use. However, the opinions of the investigators are far from concordant.

Table No. 3 gives some of these maximum quantities. The data in the last column refer to the best drinking waters of Northern Germany³.

For carrying out a complete chemical analysis of water, no less than 5 liters of it are required.

In investigating the chemical properties of water⁴, first of all the total amount of solid mineral impurities is determined—the so-called “residue after evaporation”—in milligrams per liter of water. For this purpose, 0.5–1 liter of water is evaporated in a weighed platinum dish on a water bath; after all the water has evaporated, the dish is placed in a drying oven at \(110^\circ\) C (in America evaporation is carried out at \(100^\circ\)) and weighed after three hours’ drying; then it is again placed in the oven and weighed a second time after an hour. This is continued until the difference between weighings is reduced to tenths of a milligram.

If the water was turbid, it is also necessary to determine the quantity of mechanically suspended substances. For this purpose, 0.5–1 liter of the turbid water is evaporated, the water being well shaken before measuring; alongside this, an equal volume of filtered water is evaporated. The difference between the dried residues constitutes the weight of the mechanically suspended substances.

For an approximate determination of the content of organic impurities in the residue, the residue is ignited to red heat and weighed again after cooling. On ignition the organic impurities burn, but at the same time volatile compounds are evolved, and, moreover, some of the salts decompose; for example, calcium carbonate (\(\mathrm{CaCO_3}\)) changes into calcium oxide (\(\mathrm{CaO}\), lime), losing carbonic acid (\(\mathrm{CO_2}\)). Therefore, the loss on ignition does not express the true quantity of organic substances.

For determining the total content of mineral salts, in many cases it is more convenient to use the Kohlrausch-Ostwald method, based on the increase in the electrical conductivity of water with the increase in its salt content.

The apparatus (constructed according to Pleissner’s instructions^7), see Figs. 1 and 2) consists of a glass vessel for the water being tested, into which two metal plates serving as electrodes are lowered; the distance between them remains constant. The glass vessel is placed in a vessel or box in which a definite temperature is maintained (the electrical conductivity of water depends on temperature). Then the resistance of the liquid to the passage of current through it (between the electrodes) is measured, using the alternating-current method, which produces a sound in the telephone in the event of an inequality of potentials.

Calling the sought resistance of the water \(W\), the known resistance introduced into the circuit \(R\) (the apparatus has three coils of 10, 100, and 1000 ohms), and the resistances of the segments of the graduated ruler through \(AB\) and \(BC\), we determine \(W\) from the equation

\[ \frac{R}{W}=\frac{AB}{BC}, \quad \text{whence } W=R \times \frac{BC}{AB}. \]

The salt content in the water being tested is determined from the included \(AB\) according to the table attached to the apparatus.

Hardness of water. After the total content of solid impurities in water, the so-called “hardness” of water is also of very important significance, i.e., the quantity of salts of the alkaline-earth metals contained in it (calcium, magnesium, etc.). Hardness is measured in degrees, which denote the weight content of salts causing the hardness of the water. In different countries different values are assigned to degrees of hardness; German degrees (also adopted in Russia), French degrees, and English degrees are distinguished.

One German degree denotes the content of one part by weight of lime (CaO), or an equivalent quantity of magnesium salts, in 100,000 parts of water, i.e. 0.01 gram of lime in 1 liter of water.

A French degree corresponds to the content of 1 part by weight of calcium carbonate \((\mathrm{CaCO_3})\) in 100,000 parts of water, and an English degree—to 1 part of calcium carbonate \((\mathrm{CaCO_3})\) in 70,000 parts of water.

Since 1 part by weight of \(\mathrm{CaCO_3}\) contains 0.56 parts by weight of \(\mathrm{CaO}\), then

\[ 1 \text{ French degree} = 0.56 \text{ German} = 0.70 \text{ English}. \]

\[ 1 \text{ English degree} = \frac{1}{0.70}=1.43 \text{ French} = \frac{0.56}{0.70}=0.80 \text{ German}. \]

\[ 1 \text{ German degree} = \frac{1}{0.56}=1.79 \text{ French} = \frac{1}{0.80}=1.25 \text{ English}. \]

Since the amount of carbonates of calcium and magnesium \((\mathrm{CaCO}\) and \(\mathrm{MgCO_3})\) soluble in water depends on the amount of dissolved

^7 “Wasser und Abwasser,” 1910, II, p. 249.

in it of carbonic acid ($\mathrm{CO_2}$), then, when the water is boiled, owing to the liberation of $\mathrm{CO_2}$, part of these salts precipitates, forming a sediment, and the hardness of the water decreases.

These processes proceed according to the equations:

$$ \mathrm{CaH_2(CO_3)_2 = CaCO_3 + H_2O + CO_2} $$

$$ \mathrm{MgH_2(CO_3)_2 + 2H_2O = Mg(OH)_2 + 2H_2O + 2CO_2} $$

Salts of sulfuric acid, nitric acid, and chlorine compounds, when boiled, for the most part remain in the water in dissolved form.

The hardness of water remaining after boiling is called permanent hardness, and the hardness of water before boiling—total hardness. The difference between total and permanent hardness is called temporary or precipitating hardness.

Hydrotimetry. The hardness of water can be accurately determined by means of a complete quantitative analysis; to calculate the degree of hardness, the quantity of all salts of metals contained in the water, except calcium salts, must be converted into degrees of hardness in relation to the molecular weights of lime and of these salts, and the resulting degrees of hardness added together. For the calculation one may use Table No. 4.

Table IV.

Name of compound.	Chemical designation.	1° of hardness corresponds to a content in grams of
Lime	$\mathrm{CaO}$	1.00
Magnesia	$\mathrm{MgO}$	0.714
Carbonic acid	$\mathrm{CO_2}$	0.786
Sulfuric anhydride	$\mathrm{SO_3}$	1.429
Calcium carbonate	$\mathrm{CaCO_3}$	1.76
Calcium bicarbonate	$\mathrm{CaH_2(CO_2)_2}$	2.89
Calcium sulfate (gypsum)	$\mathrm{CaSO_4}$	2.43
Magnesium carbonate	$\mathrm{MgCO_3}$	1.50
Magnesium bicarbonate	$\mathrm{MgH_2(CO_3)_2}$	2.61
Magnesium sulfate	$\mathrm{MgSO_4}$	2.14
Ferrous sulfate	$\mathrm{FeSO_4 + 7H_2O}$	4.96
Soda	$\mathrm{Na_2CO_3}$	1.89
Caustic soda	$\mathrm{NaOH}$	1.43
Sodium bisulfate	$\mathrm{NaHSO_4}$	4.28
Chlorine	$\mathrm{Cl}$	0.634

In practice, one often confines oneself to an approximate method of determining the total amount of earthy salts in water, called “hydrotimetry.” This method, proposed by Clark (England) and improved by Boutron and Boudet (France), is very simple and requires no special apparatus, except for a graduated glass tube (“hydrotimeter”).

This method is based on the property of a soap solution (in alcohol) to form, in water that contains no earthy salts (calcium or magnesium), a foam that remains on the surface; on the contrary, if these salts are present in the water, then in the presence of the fatty acids contained in soap they form insoluble compounds that prevent the appearance of foam. Foam will appear only after all the calcium and magnesium salts have been neutralized by the fatty acids of the soap. To neutralize 10° (German) of hardness in 1 liter of water, about 1.2 grams of white soda soap are required.

Thus, from the amount of titrated soap solution added to the water, one may judge the quantity of calcium and magnesium salts, and consequently (approximately) the hardness of the water.

According to Boutron and Boudet, to prepare a titrated soap solution one must dissolve 100 g of well-dried white Marseille (soda) soap in 1600 g of 90-degree alcohol, heat to boiling, and, after cooling, filter; then add 1000 g of distilled water. This gives the “first hydrotimetric solution” (liqueur), which is used for determining the hardness of water. The soap solution is poured into a burette (a narrow glass tube provided below with an opening and a stopcock for letting out the liquid in drops), on which divisions are marked so that 2.4 cubic cm of its volume are divided into 23 equal parts. The first (upper) division is not counted; 0 is placed at the second division, since for the appearance of foam it is necessary to pour in a certain excess of soap solution. Each division of such a burette (the “hydrotimeter”) corresponds to 1° of (French) hardness when the above-mentioned soap solution is used.

To verify the correctness of the composition of the soap solution, a “second solution” is used—a solution of 0.25 g of calcium chloride in 1 liter of distilled water. This latter solution, in turn, is checked by adding a solution (in \( \frac{1}{60} \)) of ammonium oxalate, of which, if the “second solution” is correctly composed, 0.126 g per 1 liter is required.

The second solution has 22° of (French) hardness. For the investigation, one always takes 40 cubic cm of the liquid whose hardness is being determined. If 40 cubic cm of the “second solution” are taken, and soap solution from the burette is gradually added to it, shaking vigorously after each addition in order to obtain foam, then foam should appear when soap solution has been added in the full volume of the burette (i.e., at 22°); if foam is obtained with some other quantity of solution added, then the solution has been prepared incorrectly and must be made anew.

After checking the soap solution, 40 cubic cm of the water under examination are taken and soap solution is added to it until foam is obtained; do—

the added quantity of solution indicates the number of degrees of hardness. The foam should be in a layer of about 5 mm and remain for 5–10 minutes.

The Boutron and Boudet method is applicable for water below 30° (French) hardness (= 16.8° German); at greater water hardness such a considerable quantity of insoluble flakes is formed that they hinder the formation of foam. Therefore, water of greater hardness should first be diluted with a definite quantity of distilled water (for example, equal to the quantity of water being tested, or twice as much); in accordance with this, the resulting degree of hardness must be increased (in the first case—by 2, in the second—by 3 times).

As indicated, 22 degrees of soap solution must neutralize 40 cu. cm. of a solution of calcium chloride containing 10 mg of this salt; consequently, 1° of solution corresponds to 0.455 mg of CaCl in a cubic cm. of water, or 11.4 mg of CaCl in a liter of water. But 11.4 mg of CaCl is equivalent, in hydrotimetric terms, to 1.03 mg of calcium carbonate (CaCO₃); thus this method determines the hardness of water in French degrees with sufficient accuracy.

Sometimes a more detailed analysis of water hardness is carried out by means of 4 successive titrations with the aid of a hydrotimeter⁵.

After determining the total hardness of the water (A°), 2 cu. cm. of ammonium oxalate solution are added to 50 cu. cm. of this water; after filtering, 40 cu. cm. of the resulting mixture are taken and its hardness B° is determined. The difference between the total hardness of the water (A°) and the hardness (B°): (A−B)° is the hardness arising from lime salts, since lime, as is known, is precipitated from solution by ammonium oxalate.

Next, 100 cu. cm. of the water being tested are boiled for half an hour in a glass flask, the loss being made up (to the original volume) with distilled water, and after cooling they are filtered. The carbonated lime precipitated during boiling remains on the filter. The hardness of the filtered water C° is the “permanent hardness.”

Finally, 50 cu. cm. of the water obtained after the preceding operation are taken and mixed with 2 cu. cm. of ammonium oxalate solution; it is allowed to settle, filtered, and the hardness D° is determined again, which expresses the magnesium salts.

Thus A° expresses the total hardness (depending on carbonic acid and salts of calcium and magnesium).

A−B corresponds to calcium salts.

C−D—to calcium sulfate.

A−B (C−D)—to calcium carbonate.

B−D—to carbonates.

The amount of calcium is determined by this method with an accuracy of up to 1–2%.

The Pignet and Hue method. To determine the hardness of water in the field, during surveys, one may use previously prepared pieces of soap, each corresponding, for example, to 4 degrees of hardness. The pieces of soap are gradually dissolved in 40 cubic cm of water, shaken, and it is observed whether foam remains. The number of pieces of soap required for foam to appear indicates the degree of hardness of the water.

Drawbacks of hard water. The suitability of hard waters for drinking depends on the composition of the salts that cause the hardness. Water containing a considerable amount of calcium carbonate is usually more suitable for consumption than water of the same hardness but containing gypsum or magnesium. The admixture of gypsum and magnesium in water is usually harmful, but only for persons unaccustomed to such water; over time the organism becomes accustomed to these admixtures.

The presence in drinking water of calcium carbonate, even in considerable quantity, generally speaking, is unquestionably harmless to humans. Thus, in the German town of Burgel (in Thuringia), about 1,000 inhabitants use water of 103° (German) hardness (of which 88° depend on the admixture of calcium, and 15° on magnesia), without any harm to health⁶.

On the contrary, it has been established that prolonged use of soft water causes, for example, diseases of the teeth, and sometimes other disorders of the organism as well⁷.

However, hard water is unsuitable for use in household and industry. Laundry cannot be washed in hard water, since soap forms insoluble flakes, and before washing the linen much soap must be expended to neutralize the earthy salts in the water. Thus, for example, to neutralize 1 cubic m of Seine water, 2 kg of soap are required. For the city of Glasgow, Parkes calculated that the annual saving in soap achieved by the city after the installation of key water amounted to 900,000 francs.

When cooked in hard water, vegetables become hardened; therefore such water is unsuitable for the kitchen. Finally, when boilers are fed with hard water, lime and magnesia are precipitated upon heating the water and form boiler scale, which damages boilers; therefore, in technical applications of water it is necessary to take care of its softening.

It is usually considered necessary to soften water if its hardness reaches 20° (German).

Significance of other impurities. The content of iron in water, in amounts above 0.9 mg per liter, although harmless to health, nevertheless

imparts to the water an unpleasant taste and the color of rust; moreover, such water strongly corrodes water-supply pipes, owing to the rapid multiplication in it of iron bacteria, which corrode iron.

For cooking and for laundry, waters containing iron and manganese are likewise unsuitable.

Lead, which usually enters water when it passes through lead pipes that distribute water inside houses, is very harmful to health; many cases of water poisoning are known. A lead content in an amount of less than 0.35 mg per 1 liter is scarcely noticeable; however, water with such a lead content should not be consumed, since it has been established¹¹ that the daily consumption of 0.55 mg of lead over the course of several months causes poisoning.

Lead dissolves in water containing free carbonic acid, or organic acids (humic and ulmic—in water passing through peat).

Copper, sometimes entering from copper pipes, and also used for the disinfection of water in the form of readily soluble salts, is harmless when its content is 10–30 mg. It can have a serious effect on health only in quantities above 200 mg.

Inorganic nitrogenous compounds entering water (ammonia, nitrites, nitrates), as well as dissolved gases and organic substances, are important not in themselves (since they are usually present in negligible quantities), but as indicators that the water containing them is suspect from the hygienic point of view, being the result of a process of decomposition.

Organic Impurities

Organic impurities in surface waters are of animal and vegetable origin; the former, decomposing, form ammonia, which upon further oxidation passes into nitrous and finally into nitric acid. This process is called the process of mineralization of organic substances.

During the decomposition of plants, various organic compounds are obtained (humic, or mouldering, substances), solid and gaseous, which do not contain ammonia. Thus, ammonia is a reliable indicator of the entry into water of products of decomposition of organic substances of animal origin.

In practice, one often confines oneself to an approximate determination of the total quantity of organic substances by the amount of oxygen required for their oxidation. To the water under investigation is added—

¹¹ Brouardel’s opinion; see Lehmann, Methoden der praktische Hygiene, 1901, p. 618.

a solution of potassium permanganate (manganese chameleon), \((\mathrm{KMnO}_4)\), which has a bright red color; giving up part of its oxygen for the oxidation of organic substances, the chameleon is converted into manganous potassium, losing its color.

Determination of oxidizability by Kübel’s method is carried out in the following manner⁸:

Into a flask of 300–400 cubic cm capacity, pour 100 cubic cm of the water being tested, 5 cubic cm of a (1:3) solution of sulfuric acid, and add from a burette with a stopcock 10 or more cubic cm of the so-called normal chameleon solution (0.316 g per 1 liter of water); each cubic centimeter of this solution corresponds to 0.08 mg of oxygen; the flask is placed on a tripod with a wire gauze and boiled for exactly 10 minutes; then 10 cubic cm of oxalic acid solution (0.63 g per 1 liter of water) are poured in, whereby the solution is decolorized; then chameleon is again added, at first in small portions and toward the end dropwise, until a persistent faint pink coloration appears, and the total amount of chameleon added is noted.

Immediately upon completion of the experiment, while the liquid is still hot, 10 cubic cm of oxalic acid solution are added and it is again titrated with chameleon as before; this second experiment shows the ratio of the chameleon to the oxalic acid solution and serves to check the correctness of the composition of the chameleon solution.

The reactions occurring in these two processes are expressed by the following formulas:

\[ 2\mathrm{KMnO}_4 + 3\mathrm{H}_2\mathrm{SO}_4 = 2\mathrm{MnSO}_4 + \mathrm{K}_2\mathrm{SO}_4 + 3\mathrm{H}_2\mathrm{O} + \mathrm{O}_5 \]

(the chameleon, combining with sulfuric acid, liberates oxygen, which goes to oxidize the organic substances)

\[ 2\mathrm{KMnO}_4 + 3\mathrm{C}_2\mathrm{H}_2\mathrm{O}_4 + 3\mathrm{H}_2\mathrm{SO}_4 = 10\mathrm{CO}_2 + 2\mathrm{MnSO}_4 + \mathrm{K}_2\mathrm{SO}_4 + 8\mathrm{H}_2\mathrm{O} \]

The amount of chameleon required for the oxidation of the water is calculated as follows:

Suppose that in the second experiment 9.9 cubic cm of chameleon were used for the oxidation of 10 cubic cm of oxalic acid; consequently, the chameleon solution is stronger than normal by \(\frac{10}{9.9}\) times. For the oxidation of the organic substances and 10 cubic cm of oxalic acid in the first experiment there were used, for example, 25.3 cubic cm; of these, for the oxidation of the organic substances there will be

\[ 25.3 - 9.9 = 15.4 \]

cubic cm of solution, which corresponds to

\[ \frac{15.4 \times 10}{9.9} = 15.56 \]

cubic cm of normal solution, or \(15.56 \times 1.16 = 4.92\) mg of chameleon.

Consequently, for the oxidation of the organic substances contained in 1 liter of water, \(4.92 \times 10 = 49.2\) mg of chameleon, or \(15.56 \times 0.08 \times 10 = 12.45\) mg of oxygen, is required.

When examining water that is very heavily contaminated with organic substances, it is first diluted with distilled water, and then the corresponding correction is made in the results obtained.

It should be borne in mind that ferrous oxide, hydrogen sulfide, and nitrous acid also decolorize potassium permanganate; therefore, when these substances are present in the water, their content must first be determined. With regard to nitrous acid, we shall point out that 1 part of it absorbs 0.42 parts of oxygen, or 1.66 parts of chameleon.

Determination of the putrescibility of water. As we have said, the presence of ammonia, nitrous and nitric acids⁹ in water indicates the results of the process of decomposition (putrefaction) of organic substances. In order to find out whether this process has ended or is still continuing in the water, one resorts to determining the putrescibility of the water. We shall briefly describe the simplest method of this determination, adopted in England, the so-called “incubator test.”

Having determined the quantity of oxygen absorbed by the water under examination over the course of 3 minutes, the water is kept for 5 days in a water bath at \(27^\circ\), and then the oxidizability over 3 minutes is again determined. If putrefaction has occurred during this time, the quantity of oxygen absorbed should increase considerably, since the products of decomposition oxidize much more rapidly. If, on the contrary, the water has kept well, the quantity of oxygen absorbed in 3 minutes will decrease. Some investigators perform this test by keeping the water at \(18^\circ\) for 7 days.

In carrying out a chemical analysis, it is very important to interpret its results correctly.

Knowing the quantity of the individual elements contained in the water is less important than knowing their origin and, consequently, the changes to which the water has been subjected in the different parts of its course. Thus, for example, a considerable content of compounds of chlorine, sulfates, and phosphates has no bad significance if it is explained by the geological composition of the strata through which the water has flowed; but the quantity of these impurities has an entirely different significance if it is proved that they have entered the water not from the soil, but only by way of, or past, it and, consequently, are the result of direct or indirect contamination of the water. In exactly the same way, a considerable admixture of organi-

…substances of vegetable origin are far less harmless than an insignificant admixture of animal origin.

For a correct assessment of the water of any source, its composition must be compared with the “normal water” of that same source, i.e., with the sort of water that the given source would have (depending on its location, the geological structure of the locality, etc., under natural conditions) if it had been left in its natural state and had not been subjected to any artificial, extraneous pollution.

Biological Properties of Water

We have seen that water may contain, in suspended form, earthy particles, chiefly siliceous and clayey, more or less fine; but it often also contains other bodies, visible or invisible to the unaided eye, belonging to the vegetable or animal kingdom. Sometimes living organisms get into it, and sometimes parts of dead organisms. These organisms, especially if they are still alive and capable of reproducing and leading a parasitic mode of life, are, of all admixtures in water, of the greatest importance for the health of human beings and animals using the water; moreover, their presence in water serves as an indicator of those phenomena that occur in the water and in which these microorganisms take part. It may be said that the microorganisms found in water are the more important the smaller they are.

The study of the microorganisms of water is the task of the natural scientist or the physician; however, with the present development of the doctrine of microorganisms, to which the foremost place belongs in the processes of pollution and purification of water, the engineer engaged in water supply must have an understanding of the microbiology of water in order to judge correctly its qualities and the means of purifying it.

Although in nature there is no sharp boundary, it is nevertheless customary to distinguish from the flora and fauna of water a group of lower organisms—bacteria, or microbes; they are especially important for the hygienist, because all causative agents of diseases belong to their number, occurring among higher microorganisms only as an exception.

Therefore the biological analysis of water may be divided into micrographic analysis (or microscopic-biological), i.e., the study (under the microscope) of the micro-flora and micro-fauna of higher order, and bacteriological analysis, or the detection of bacteria (by creating artificial conditions favorable to the development of bacterial germs).

Let us touch briefly on biological analysis.

The aggregate of microorganisms, animal and vegetable, floating in water in suspended form, is called plankton.

The quantity of plankton varies considerably in different cases: many spring and ground waters are entirely free of living organisms, even bacteria, whereas some surface waters, especially stagnant ones, contain several thousand microscopic animals and plants and several million bacteria in 1 cubic centimeter.

The development of microorganisms in water is always subject to the law according to which the character and quantity of the elements composing the plankton depend chiefly on the more or less favorable conditions presented by the water under examination for different species, and change in time and space only in accordance with changes in these conditions.

In other words, each species of microorganism develops in water only if it finds in it the physical conditions suitable for itself (heat, light, rest or motion, etc.), the chemical composition (oxygen, nitrates, albuminoids, etc.), and finally a supply of food (usually one species feeds on others, lower ones). When these conditions change—when the average temperature of the water falls or rises, or the oxygen content decreases, etc.—this species of microorganism immediately disappears, or, more precisely, only a few isolated specimens of it remain, ready to begin multiplying again as soon as the water once more presents favorable conditions. This species is replaced in the water by another, which has, as it were, been awaiting favorable conditions for its own reproduction. Thus microorganisms are very sensitive to changes in the properties of water, and moreover to those properties that depend on the average composition of the water, upon which accidental impurities entering the water and accidental changes in its quality have no influence.

Therefore, conversely, having studied the habits and requirements of different species of microorganisms, one may, from the presence in the water of one or another species, or of a combination of several species, easily judge the origin and composition of the water, and the nature of the contaminations that have entered it, etc.

At the present time, biological examination of water is used not only to determine the presence of microorganisms harmful to human health, but also to clarify the origin of elements detected by chemical analysis, whose appearance and transformations in water often cannot be explained by the data of chemistry. In addition to the speed and simplicity with which results are obtained (in comparison with chemical and bacteriological analyses, which require both time and a special laboratory), microscopic-biological examination has the further advantage that the data obtained by it relate, as mentioned, to the average composition of the water characteristic of the body of water under study, whereas in the taking of samples

for laboratory analysis we always run the risk of obtaining a sample dependent on the accidental admixtures present in the water at the given time.

The foundations of the biological investigation of water according to its flora and fauna have been developed only in the last ten to fifteen years,^14) chiefly by Dr. Kolkwitz (in Berlin^15)). This method is especially important for investigating the degree of pollution of rivers.^16)

In Russia a biological investigation was carried out on a large scale on the Moskva River by Dr. Nikitinsky.^17)

The properties of the species living in water are still far from having been studied; the biological investigation of water is further complicated by the fact that the microorganisms inhabiting the waters of different countries differ, depending on climate, soil, etc.; therefore, for the correctness of the conclusions drawn from the investigation, a thorough acquaintance with local conditions and an elucidation of the so-called “normal” state of a given reservoir and of the microflora and fauna normal for it are necessary.

However, the foundations for biological investigations have already been established; the properties of a whole series of microorganisms have been clarified, remaining almost unchanged under all local conditions.^18)

All species, animal and plant, are divided into two large groups: species that occur in clean water, which have been given the name “catarobic” microorganisms, and species requiring the presence of decomposition products, or “saprobic” ones. Saprobic animal microorganisms are called “saprozooa,” and plant ones—“saprophytes.” Organisms living in waters strongly polluted with products of decomposition are called “polysaprobic”; those in slightly polluted waters, “oligosaprobic”; those requiring a medium degree of pollution, “mesosaprobic.”

Let us say a few words about the composition of the plankton of fresh waters. First of all, in living plankton one must be able to distinguish certain non-living elements, or the so-called pseudo-plankton. To it belong:

1) Remains of organisms of plant origin. They are of little significance for human health, and serve only as indicators of the presence in the water of rotting wood or substances that have come from

^14) Prof. Metz. Microscopische Wasser analyse. 1898.

^15) See a series of articles on this question by Kolkwitz and Marsson in “Mitteilungen aus der Kngl. Prüfungsanstalt für Wosserversorgung und Abwasserbeseitigung” from 1902 to the present time.

^16) This question also has to be dealt with in studying the conditions of discharge of wastewater into rivers and in choosing the degree of purification of wastewater before its discharge.

^17) Nikitinsky. Second report of the Commission for the investigation of the results of biological purification of wastewater on the irrigation fields of the city of Moscow. 1909–1910.

^18) A description of the principal microorganisms, as well as of the basic methods of biological investigation, is the subject of a special appendix to the present work.

the intestines of humans or animals (boiled grains, beans, the skins of fruits or vegetables, etc.).

2) Remains of organisms of animal origin.

In waters inhabited by numerous animal organisms, naturally, the bodies, or parts of the bodies, of dead animals are encountered; some of them decompose rapidly, others slowly. These bodies are significant only as an indication of the presence of living microorganisms of the same kind.

Muscle fibers of meat (Fig. 3), droplets of fat, and pieces of intestinal epithelium indicate contamination of the water by fecal substances.

Fibers of fabric, hairs of people and animals indicate that the water has been used for washing, etc.

3) Eggs and larvae of parasitic worms (for example, tapeworms, etc.) are dangerous for humans, since they may develop in the human intestine; moreover, their presence in water indicates contamination by feces, since these eggs could have entered almost exclusively from the intestine and from excreta.

4) The transmission of diseases by animal microorganisms (protozoa) through water apparently almost never occurs¹⁰. Only Amoeba dysenteriae may possibly be transmitted by water.

5) Flora of polluted waters. Aquatic plants play a major role in the processes of natural purification and oxidation of waters, since, by the very manner of their nutrition, with the aid of chlorophyll, they assimilate carbon from carbohydrates and carbon dioxide, restoring oxygen. This process of oxygen restoration is apparently stronger in lower-class plants than in higher ones, so that the most energetic oxidation is caused by algae. Algae also destroy bacteria. Fig. 4 shows the most widespread kinds of algae.

Let us also note four species from the class Schyzophyceae (the group bacteriacea), which are of great importance for the engineer: 1) Crenothrix (Fig. 5) and 2) Leptothrix have the ability to extract from water the iron dissolved in it. They cannot develop in water without the presence of iron in it, but a content of 3 mgr. of iron per 1 liter of water is sufficient for Crenothrix and Leptothrix to develop in large quantity. The cells of these plants separate iron from the water and oxidize it; the roots, growing, form a network of ferruginous substance that adheres to the walls of water-supply pipes and constricts the cross-section of the pipes.

Under favorable conditions, Crenothrix reduces the carrying capacity of pipes by 1% in the course of a year. Owing to the presence of Crenothrix, water becomes turbid and sometimes foul-smelling.

Fig. 1.

Fig. 3.

Fig. 2.

3) Cladothrix (Fig. 6) exhibits similar activity with respect to lime salts, depositing them in enormous quantity over the filamentous branches of the plant itself.

4) Beggiatoa (Fig. 7) grow in waters with considerable organic impurities, which they process by decomposing sulfurous-acid compounds, at the same time releasing hydrogen sulfide and retaining part of the sulfur in their protoplasm.

We shall not dwell on the methods of biological analysis¹¹.

Bacteriological Properties of Water

Bacteria are divided into pathogenic, or disease-producing, bacteria, which are the causative agents of diseases, and indifferent bacteria, which have no effect on the health of humans and animals. In considering the bacterial population of water, one may also distinguish suspicious bacteria, meaning by this bacteria which, generally speaking, are harmless, but can exist only in an environment where decomposition of organic substances is taking place and therefore indicate these processes in the water; among the suspicious bacteria should be included Bacillus coli communis (the intestinal bacillus), which is always present in the human intestine but can enter water only with fecal matter. Fig. 8h shows a culture of B. coli on gelatin on the third day, and g on agar-agar (jelly made from algae).

Special attention is paid to the intestinal bacillus in the bacteriological examination of water also because the conditions of its development, as well as the methods of its determination, are almost identical with those for the typhoid bacillus (pathogenic).

Very diverse species of bacteria occur in water, both aerobic (requiring the presence of air for their existence) and anaerobic, living in places to which fresh air has no access.

In the examination of water, determining all species of bacteria present in it would be exceedingly difficult, and there is no need for it. In practice, in the bacteriological analysis of water, two requirements are set:

1) To determine whether the water contains pathogenic or suspicious bacteria, the presence of which makes the water directly and immediately dangerous for use. This analysis, by analogy with the terms of chemistry, is called “qualitative.”

MELOSIRA          FRAGILARIA

Fig. 4.

Fig. 5–7.

Fig. 8.

Fig. 9.

In practice, in the continuous control analyses of drinking water in existing water-supply systems, in cases where there is no reason to fear the appearance of pathogenic bacteria in the water, the water is tested for the presence of Bacillus Coli.

2) To ascertain to what degree the water is accessible to bacteria, by determining for this purpose the total number of bacteria of every kind present in the water; this number serves as a measure of the purity of the water, indicating its accessibility, among others, to pathogenic bacteria.

This analysis is called “quantitative.”

Concept of the Methods of Bacteriological Analysis.

Samples for bacteriological analysis must be kept in hermetically sealed vessels. For taking samples, a sterilized vessel is used, fitted with an airtight stopper that can be opened and closed again at a certain depth. Glass bulbs with a narrow neck, previously freed of air and sealed (see Fig. 9), are also sometimes used; having lowered it to the depth from which the sample is to be taken, the neck is broken off under water by a blow from the movable weight C, which is then stopped by the catch c, and water enters the bulb. After raising the bulb, it is immediately sealed.

To determine the number of bacteria (germs), they are given the opportunity to develop in a nutrient medium. Such a medium may be, for example, a boiled mixture of 6 g of Libikhov’s¹ meat broth with 150 g of gelatin (fish glue) and 1,000 grams of distilled water. This composition melts at 30°–40°. Having taken into a test tube (of course, sterilized) about 20 cubic cm of nutrient gelatin, the water under investigation is added drop by drop, thoroughly shaking the gelatin after each drop; it is dripped from a pipette that gives drops with a volume of 1/20 cubic cm. One cubic cm or less of the water under investigation is “sown,” depending on the presumed bacterial content in it, since it is desirable that no more than 100–200 bacteria be sown. The inoculated gelatin is then poured into a flat, sterilized glass dish about 8 cm in diameter (“Petri dish”) (see Fig. 10), and the dish is kept covered at a constant temperature (usually 20°–22° C). In this way the bacteria develop, and in the gelatin there appear “bacterial colonies” visible to the eye (see Fig. 11). It is assumed that each colony has developed from a single germ. After a definite number of hours (usually 48 hours), the number of colonies is counted with the aid of a magnifying glass, for convenience of counting placing beneath the bottom of the dish a black plate ruled into sectors.

Fig. 10—11.

Altona

Hambourg

Elbe Fleuve

Fig. 12.

In the qualitative analysis of water for the presence of some particular species of bacteria, inoculation is made into a medium in which only that species can develop, while other bacteria do not develop. Thus, for each species (sometimes for a group of species) of bacteria a special inoculation is required.

To ascertain the presence of Bacillus coli and, at the same time, Bacillus typhi, which in its conditions of development is very close to the former, use is made of two of their characteristics: first, these two species easily tolerate a temperature of 45°, at which most other species of bacteria inhabiting water can no longer grow; second, weak doses of certain antiseptics, namely phenolic acid, do not hinder the growth of these two bacteria, while completely arresting the development of all the others. Therefore, by adding to nutrient gelatin a certain quantity of a 5% solution of phenolic acid, we obtain in it only colonies of bacillus coli and bacillus typhi.

Among the various methods for determining the intestinal bacillus, we shall indicate the method of Péré¹², in which a large quantity of water is examined, which is sometimes important when the bacterial content is low. A mixture is taken in the following proportion: 100 cubic cm of sterilized broth, 50 cubic cm of a 10% solution (neutralized and sterilized) of pure peptone, 20 cubic cm of a 5% solution of phenolic acid, and 830 cubic cm of the water under investigation (1 gram of phenolic acid per 1 liter of mixture). The liquid is poured into 10 sterilized flasks and, having been closed with cotton wool, is kept at 34°; if B. coli or B. typhi is present, the solution becomes turbid, and the sooner so the greater the number of bacilli.

When the mixture has become turbid, an inoculation of a definite volume of it is made onto one or another nutrient composition (for example, in 10 Petri dishes), and the resulting colonies are counted.

In the case of detecting B. coli, in more recent times it has been customary to establish the so-called “coli titer,” i.e. to determine the smallest quantity of water in which B. coli is still found; for this purpose inoculations are made, for example, of 500 cubic cm of water, 300, 100, 10, 1 cubic cm, 0.1 cubic cm, and so on, in separate flasks, and it is observed which of them become turbid. (In London, where the requirements for drinking water are very high, water is considered pure if half the samples of 100 cubic cm each do not contain B. coli.)

We cannot dwell on the details of bacteriological analysis.

Among the various properties of bacteria, let us note the so-called “resistance,” i.e. the capacity to withstand the action of various bactericidal (sterilizing) agents. The resistance of different species varies; the “sporogenous” forms of bacteria are distinguished by particular resistance.

(reproducing by spores), which for the most part are harmless to humans. Thanks to this, water can be disinfected, i.e., pathogenic bacteria in it can be killed, but without achieving its complete sterilization, i.e., the destruction of all kinds of bacteria, including resistant species of harmless bacteria.

The number of bacteria in water depends on many circumstances, and everything said above about microorganisms in general applies to them as well (on the question of the biological examination of water ²²).

Let us add, moreover, that when the temperature is present at which bacteria can develop, the most important factor for them (according to Rubner’s experiments ²³) is the presence of nutrients. At the same time bacteria show the greatest vital force during the first two days; if the development of a bacterial culture is observed, it will be seen that after several hours their number begins to increase rapidly and reaches a maximum after 24–48 hours; after this their number remains for a long time without change: an equilibrium sets in, in which the number of newly originating bacteria is equal to the number dying. The development of bacteria in water is also hindered by the struggle among the various species of bacteria, the consumption of bacteria by lower plants and animals, the action of light, which kills bacteria, their settling, and so forth. Therefore bacteria live and develop in natural water only during certain periods.

Pathogenic Bacteria.

The number of species of pathogenic bacteria which in practice are transmitted by water is, generally speaking, not large. They all belong to those species which transmit infection through the human alimentary canal; these include the causative agents of cholera (Vibrio Cholerae Asiaticae), typhus, paratyphus, and dysentery.

In addition, trachoma, gonorrhea, and wound infections may be transmitted through washing with water.

Animals too may become infected through water with certain infectious diseases.

The majority of pathogenic bacteria, apparently, under ordinary conditions cannot multiply, but only persist ²⁴); the exception, perhaps, is the cholera vibrio.

²²) According to the number of bacteria in water, Miquel divides natural waters into six classes:

1) extremely clean water, with the number of bacteria in 1 cubic cm from 0 to 10.
2) “ very clean . . . . . . from 10 to 100.
3) “ clean . . . . . . . . “ 100 to 1000.
4) “ moderately clean . . . “ 1000 to 10000.
5) “ unclean . . . . . . . “ 10000 to 100000.
6) “ very polluted . . . . over 100000.

²³) Rubner, Archif für Hygiene. 1906, 57, p. 161.

According to the studies of Gotschlich¹³, cholera vibrios survive in water for up to 3 months, and typhoid bacilli for up to 4 weeks; in cold water and in ice the latter survive for up to 100 days.

The Influence of Drinking Water on the Spread of Epidemics and on Population Mortality.

The chief epidemic diseases spread by poor-quality drinking water are cholera and typhoid (abdominal typhus).

The influence of water on cholera morbidity is easier to establish than its influence on infection with typhoid, since the characteristic signs of cholera appear as early as the third day after infection, and analysis of the water source used for drinking can reveal whether vibrios are present in it; typhoid, however, usually appears only after 14–21 days, and by that time typhoid bacilli, if they happened to enter the water (together with the excreta of some patient), may already have disappeared from the water of the given source.

Cholera epidemics arising as a result of contamination with vibrios of a source of central water supply have the character of an outbreak, i.e., they suddenly appear at once in all places using the water of that source. Cholera epidemics occur mainly when insufficiently purified river water is used for drinking.

A classic example of a cholera epidemic caused by drinking water is the epidemic of 1892 in Hamburg, where during the summer of 1892 there were 18,000 cases and 8,200 deaths (14.22 deaths per 1,000 inhabitants). Hamburg used unfiltered water from the Elbe River (settled only in sedimentation basins); although the water was taken upstream of the city, the influence of the tides extends above the water-intake point, so that the city itself contaminated its drinking water with its own sewage. At the same time the city of Altona, adjoining Hamburg and lying below it along the course of the Elbe, suffered only slightly (2.42 deaths per 1,000), since it drank filtered water. In Fig. 12, dots mark the houses where there were cases of cholera: a group of houses in Hamburg affected by cholera used water from the Altona water supply.

We need not say much about the cholera epidemic of 1908–9 in St. Petersburg, where the intake pipes of the municipal water supply are located next to the place where sewers discharge into the Neva from certain buildings, for example, from the municipal hospital, in which there were several dozen cases of cholera.

Effect of the introduction of tap water (in 1873) on mortality from typhoid in Vienna.

Number of deaths from typhoid.

Legend:

• Diagonally hatched — absolute numbers
• Cross-hatched — per 10,000 inhabitants

Vertical annotation: tap water introduced

Fig. 13.

Fig. 14. Variations in mortality from typhoid in France.

— fluctuations in water consumption.
- - - fluctuations in level.

Vertical scale: In fractions of the average monthly value.

Months: I, II, III, IV, V, VI, VII, VIII, IX, X, XI, XII, I, II, III, IV, V, VI, VII, VIII, IX, X, XI, XII, I.

In all, the number of cases in 1908 (from August through December inclusive) was about 9,000, and the number of deaths was 4,000.

All investigators of cholera agree that water plays the principal role in its transmission. As regards the spread of typhoid, opinions differ: some specialists suppose that the chief mode of transmission of typhoid is contact with the sick, but they too do not deny the role of water. The considerable decrease in cases of typhoid and in mortality from typhoid (together with a decrease in general mortality) that occurs in all cities when their water supply is improved clearly indicates a connection between these phenomena.

Fig. 13 shows the change in the number of deaths from typhoid in Vienna from 1851 to 1910. Until 1873 Vienna used unsatisfactory water from the “Kaiser Ferdinands” waterworks (groundwater), and in 1874 water began to be taken from springs (Kaiser Frans-Joseph. Hochquellenleitung).

Table No. 5 gives figures for the decrease in mortality in certain cities of America and Western Europe after the reconstruction of their water supply.

From the comparative graph (Fig. 14) of fluctuations in water consumption by month in the cities of France (shown by a solid line) and fluctuations in mortality from typhoid (shown by a dotted line), it is evident that both curves are very similar to one another, but the mortality curve lags behind the water-consumption curve by approximately one and a half months (the maximum of mortality falls in September, while the maximum of water consumption is in July and August). This lag is quite understandable: death from typhoid does not occur immediately after the infection of the organism by typhoid bacilli; typhoid has an incubation period (i.e., the interval of time between infection of the organism and the first manifestations of the disease) of not less than 2–3 weeks, and the illness itself also lasts several weeks. On the mortality curve (Fig. 14) we see only one sharp deviation from the water-consumption curve: an increase in mortality in March, which does not depend on the quantity of water consumed, but is explained both by the general increase in mortality (i.e., from all sorts of diseases) in spring and by the deterioration of water quality at this time of year (owing to the melting of snow and the increase in the velocity of river currents, whereby the water is stirred up or carries from the river bottom material containing many bacteria).

The graph in Fig. 14 was compiled from average figures for 50 cities in France over 5 years (1899–1903).

The fluctuations in the number of cases of typhoid in individual cities—Breslau, Berlin, Hamburg, and Marseille—shown in Fig. 15 are similar to the fluctuations in the graph of Fig. 14. All this confirms the theory of the spread of typhoid by drinking water (“Koch’s theory”).

Table V.

City name	Reorganization of the water supply (replacement of supply with unfiltered river or well water)	Year of reorganization	Typhoid mortality per 1,000 inhabitants (average over 5 years): before reorganization	Typhoid mortality per 1,000 inhabitants (average over 5 years): after reorganization	Overall mortality per 1,000 inhabitants (average over 5 years): before reorganization	Overall mortality per 1,000 inhabitants (average over 5 years): after reorganization
Lawrence	Filtration.	1893	1.21	0.26	24.4	20.0
Albany	—	1899	1.04	0.28	22.4	18.4
Lowell	Underground springs.	1896	0.97	0.21	25.1	20.5
Newark	Mountain springs.	1892	0.70	0.16	25.1	22.1
Vienna	Mountain springs.	1873	1.05	0.52	39.4	30.6
Zurich	Filtered lake water.	1885	0.76	0.10	—	—
Hamburg	Filtration of water from the Elbe River.	1893	0.47	0.07	24.0	17.7
Nancy	Filtering water-collecting gallery.	1882	1.06	0.72	25.8	23.6

Fig. 15.

Fig. 16.

According to another theory, belonging to Pettenkofer and developed by Emmerich, epidemics of typhoid and cholera can spread only on porous soil permeable to air and water, and chiefly when the soil has a definite, relatively insignificant moisture content, i.e., when the groundwater level is low.

Indeed, in many cities a low level of groundwater coincided with an increase in mortality from typhoid, so that on this basis one may assume a dependence between these phenomena. This coincidence was especially characteristic in Munich in the period when Munich was still subject to typhoid epidemics (before 1880) (see Fig. 16; ²⁶). However, at the present time Koch’s theory predominates.²⁷

Typhoid bacilli can be transmitted not only by drinking water, but also through the consumption of (raw) vegetables, fruits, etc., washed with water; cases are also known in which epidemics were transmitted by oysters grown in infected water.

Requirements Imposed on Drinking Water

The requirements imposed on drinking water have by no means been the same at different times and among different peoples.

The essence of these requirements, among all peoples, consciously or instinctively, comes down to one thing: drinking water must not harm health. However, notions of “harmlessness to health” change greatly, and the requirements for ensuring safety to health rise along with the development of hygiene and the raising of its standards. Moreover, the understanding of the various processes taking place in water, and of the role of its various admixtures, organic and inorganic, also develops and sometimes changes fundamentally, in connection with advances in chemistry, bacteriology, and other branches of knowledge, whose conclusions are used by hygienists.

²⁶ In Figure 16, solid black shading shows the absolute number of deaths from typhoid in Munich, while hatching shows the position of the groundwater level; moreover, the lower points in the drawing correspond to the highest level, and the upper points to the lowest position of the water level.

²⁷ If, at times, the installation of proper water supply does not reduce mortality from typhoid (as was the case, for example, in Danzig, Munich, and Berlin), this is explained by the absence of sewerage and the necessity of installing it.

According to French legislation (law of February 15, 1902, Article 9), if in any commune, over the course of three consecutive years, the mortality figure exceeds the average mortality figure of France (at present about 22 persons), measures must be taken for the improvement of this commune, with the installation, if necessary, of proper water supply and sewerage, if these did not yet exist.

Views on drinking water may, from the hygienic point of view, be divided into three periods:

The first period, from the most ancient times to the middle of the nineteenth century (approximately to 1850), when the basis for judging water with respect to its suitability for drinking consisted chiefly of its physical properties.

The second period (1850–1880), when, in the hygienic evaluation of water, alongside its physical properties, chief and decisive importance was attached to its chemical composition, namely to the presence in it of foreign substances not belonging to its usual composition.

The third period, which began around 1880 and continues to the present day, representing the epoch of the emergence and development of a new science—bacteriology—is distinguished by the fact that, in the hygienic evaluation of water, alongside its physical and chemical properties, the bacteriological properties of water play an outstanding role.

It must be assumed that the “bacteriological” period in the evaluation of water will continue for a long time yet: bacteriology is still far from having spoken its final word; views on the transmission of infectious diseases by pathogenic bacteria, on the spread of these bacteria in the soil, on the antagonism (struggle between species) of bacteria, and so forth, cannot yet be regarded as firmly established; meanwhile the steady advances of bacteriology allow one to hope that all these questions will find a clear resolution.

It is difficult to say what new paths the science of human health will open up, and whether a fourth period will arrive in views on the properties of water, when to the requirements of physical, chemical, and bacteriological purity some other new requirements will be added. It may be that a new period will arise in modern times in connection with the development of the study of the radioactive properties of various bodies.

Thus, already now attempts are being made^28) to explain by radioactivity the unexplained property of the water of certain springs to cause goiter, which was formerly groundlessly attributed to the hardness of the water^29).

The practical requirements imposed on drinking water are also increasing in connection with advances in water-purification technology.

In recent years, large competitions for the purification of drinking water have been held in Paris (1905–1908) and Marseilles (1910), and at the present time competitive trials of various water-purification systems are being carried out in St. Petersburg.

^28) Repin. Comptes rendus des séances de l’Académie des sciences, 1908, 147, p. 387.

^29) Among waters leading to the development of goiter, there are known, for example, the waters of the springs Willard-Clément and St. Pancrace near St. Jean de Maurienne in Savoy.

The successes of the technique of purifying drinking water, as revealed at the aforementioned competitions, make it possible to raise the sanitary requirements imposed upon drinking water supplied by central water-supply systems, without thereby significantly increasing the charge for water. The latest requirements may be stated in the following form¹⁴:

1) Physical and chemical properties.

a) Water must always be clear and, as far as possible, colorless, without any foreign odor or taste;

b) water must not contain any substances alien to its normal composition which might have even the slightest harmful effect on the health of the population, or interfere with the use of the water for cooking, in domestic life or in industry, or, finally, lead to the destruction or clogging of the network, reservoirs, and service pipes.

2) Bacteriological properties.

c) Purified water must be freed from pathogenic microorganisms, such as, for example: from the bacteria of typhus and paratyphus, tuberculosis, cholera vibrios, and putrefactive bacteria (Bacterium pyocyaneum, staphylococcus, streptococcus, etc.).

At the present time it is recognized that water which does not contain the intestinal bacillus (B. Coli) in samples of sufficient size (100–400 cubic cm), taken at the outlet from the purification devices, is sufficiently guaranteed against the presence in it of pathogenic microorganisms.

d) In purified water, on leaving the filters, only the presence of a small number (not more than 100 bacteria in 1 cubic cm) of harmless forms may be permitted, especially resistant ones, e.g., spore-bearing forms: (B. subtilis, B. mesentericus), or molds and yeasts. In this case the count must be made by inoculating no less than 1 cubic cm of the water under investigation.

3) Micrographic properties.

e) Purified water must be freed from living organisms, as well as from the eggs, larvae, and embryos of animal parasites.

Modern methods of purification and disinfection of water fully permit the fulfillment of the above requirements.

CHAPTER II.

Classification of Methods of Water Purification.

Methods of purifying water and improving its quality may pursue the following aims:

1) Removal, or reduction of the content, of floating insoluble (“suspended”) particles.

2) Change in the content in water of certain dissolved substances, i.e., improvement of the chemical properties of the water.

3) Destruction of living organisms present in the water, either all of them (sterilization of water) or only those dangerous to health (disinfection of water).

4) Improvement of the temperature of the water.

According to the techniques employed for improving water, methods of water purification are: mechanical, physical, chemical, and mixed.

For the removal of suspended substances, mechanical processes are usually employed: straining through a screen to free the water from coarse impurities; settling, for the precipitation of substances heavier than water; and filtration, i.e., passage through various porous substances (sand, artificial porous stones, etc.), which retain a large part of the suspended impurities, including microorganisms. As we shall see below, filtration cannot be regarded as a purely mechanical process; biological factors also take part in it.

In some cases, for better removal of suspended substances, chemical reagents (“coagulants”) are added to the water, making it possible to carry out the process of filtration considerably faster than without the addition of reagents; the operation of rapid (American) sand filters is based on this.

Among dissolved substances, it is especially important to remove lime from water (“softening” of water), iron and manganese; to reduce the content of acids that corrode lead pipes; and to freshen saline water.

Finally, for the disinfection of water, at the present time methods are used that are both physical (boiling and distillation of water, the action of ultraviolet rays) and chemical (treatment of water with chlorine compounds and other reagents, and ozonation of water).

We have not yet mentioned improvement of the physical properties of water, except for temperature. Improvement of the physical properties is achieved by the same methods that have already been enumerated by us.

To improve the taste of water there is no need to take special measures; if an unpleasant aftertaste depends on chemical impurities (a salty taste, an aftertaste of rust), then it will be removed by freshening or deferrization of the water, while other aftertastes will disappear with the general purification of the water.

The odor of water is caused either by gases contained in it, which are removed by airing (by aeration of the water), or by microorganisms, living or dead (chiefly algae), multiplying in lakes and reservoirs and on sand filters; to remove such an odor it is necessary to combat the development of these microorganisms. Measures against the development of algae are the protection of basins from access of light, in the absence of which most algae cannot develop, and the addition to basins containing water of a small quantity of copper sulfate (from $\frac{1}{10.000000}$ to $\frac{1}{100.000}$): so insignificant a content of copper is harmless to human health, but acts destructively on algae¹.

The turbidity of water disappears when suspended substances are removed; coloration, however, when not inherent in the water itself, depends on its chemical composition and usually arises from organic substances that have long been in contact with the water; river water most often possesses coloration. To eliminate or weaken the coloration of water, the riverbed should be cleaned (if, of course, the coloration is not caused by the properties of the soil itself, e.g., peaty soil); decolorization of water is achieved by removing the impurities that color it (by aeration or settling and filtration with the addition of aluminum sulfate).

Improvement of the temperature of water is not applicable on a large scale and is never carried out in central water supplies; the only measure that should be borne in mind when designing water-supply structures of every kind consists in preserving the temperature possessed by the water entering from the source of supply; this aim is most often achieved by burying reservoirs, the pipe network, etc., to a sufficient depth in the ground.

Cooling water in small quantities, in the absence of ice, is most simply carried out by means of evaporation, during which heat is absorbed; in warm countries, for this purpose, water is kept in vessels made of lightly fired unglazed clay; part of the water seeps through the walls and evaporates from the outer surface

them. A necessary condition for successful cooling is a low relative humidity of the air and sufficient movement of it.

In considering the various methods of water purification, we shall try to give indications regarding the choice of one or another method of purification in each particular case.

REMOVAL OF SUSPENDED MATTER.

Straining of Water.

Before entering the pipes that take water from the source of supply, the water is usually passed through gratings or screens of various construction, which retain large objects floating in the water.

Sometimes, however, water is strained through screens even immediately before filtration (or sedimentation), in order to retain those suspended and floating objects which are still present in the water, and to facilitate the subsequent purification of the water.

The structures in which the devices for straining water are placed are called “screening houses.”

Usually, when water is strained before its further purification, this is arranged together with the “aeration” (airing) of the water, i.e., mixing it with air.

An example of such an arrangement is the screening house in St. Petersburg (Fig. 17). Water, through a pressure pipe, enters the trough C, of rectangular cross-section, riveted from iron. In the longitudinal walls of the trough narrow slots \(e\) are arranged, closed by special slide gates provided for this purpose with rods having screw threads; the handles of the rods are located above the platform with which the trough is covered. Along the side walls of the trough, beneath the openings \(e\), inclined copper screens \(f\) are arranged, having 600–900 openings per 1 sq. inch; the screens are stretched on iron frames and rest on the walls of concrete troughs \(g, g\). The water, pouring in a flat jet onto the screens, is broken up into small streams having a large surface of contact with the air; at the same time the gases contained in the water are released, and the water is saturated with atmospheric oxygen. At the same time, coarse impurities are retained on the screens\(^{1}\), and are removed partly by the force of the water jet, partly by special scrapers, into the troughs \(k, k\), and from them are carted away to a dump.

The screens \(f, f\) are arranged inclined to the horizon at an angle of 10–25°, in order to make the water, spreading over the screen, break up into a greater number of streams; at the same time, with a greater inclination of the screens, the refuse retained on them is more easily washed off into the troughs \(k, k\).

\(^{1}\) Among which, in Petersburg, small fish are often found.

From concrete troughs g, g the water enters a reservoir, from which it is conveyed by pipes to the place of further purification (to the filters).

Settling of Water

Particles of those bodies whose specific gravity is greater than the specific gravity of water may nevertheless not sink in water, but float (remain in suspension), if the water is in motion; moreover, the greater the velocity of the water’s motion, the larger the floating particles of a given substance may be. As the velocity of the water flow decreases, the impurities suspended in it settle: first the larger and heavier particles fall to the bottom, then, with a further decrease in velocity, the smaller and lighter ones. Thus, the settling of turbidity from water can be achieved by leaving the water for some time in a state of rest, or by causing it to move with a definite, very low velocity.

In nature we see the process of settling in lakes.

Reservoirs in which the settling of water is carried out in a state of rest are called settling reservoirs of intermittent action; water is admitted into such a reservoir, remains in it in complete rest for a definite interval of time, and then the upper, clarified layers of water are discharged from the reservoir.

In settling reservoirs of another type, “continuous-action” reservoirs, the water moves very slowly but continuously through the reservoir, so that each newly entering particle of water travels in the reservoir over a path of definite length and with a definite velocity (and consequently remains in the reservoir for a definite interval of time). The settling of turbidity likewise takes place continuously and, with a constant composition of the water, uniformly.

At the present time, settling reservoirs of continuous action are installed almost exclusively; their principal advantage is the simplicity of supervising them, since no constant observation of the admission and discharge of water is required; furthermore, in intermittent settling the time of admitting and discharging the water is lost for the process of settling, and, finally, during this time part of the sediment that had previously settled is set in motion in them.

In order to determine the dimensions of settling reservoirs, it is important to ascertain by experiment at what velocity the turbidity in the water being clarified settles in the best manner, and what path length is sufficient for the clarification of the water.

The process of settling of suspended substances was studied in considerable detail by Steuernagel² (in Cologne) for wastewater; general conclusions—

its water may also be applied to the sedimentation of drinking water, although, of course, the figures will change. The experiments were carried out at velocities from 4 to 77 millimeters per second, and showed, first of all, that at low velocities of flow of the water being clarified, the greater part of the suspended matter settles closer to the beginning of the reservoir (i.e., to the place where water enters it); whereas when the velocity is increased, the deposits are distributed more uniformly over the entire bottom of the reservoir.¹⁵ From this Steuernagel concludes that at low velocities sufficiently short basins are adequate, while higher velocities necessitate the construction of longer basins.

Further, when settling at rest, after 1 hour of settling about 65% of the suspended matter had settled out; after a 3-hour stay of the water in the settling tank this amount increased to 70%, and after a 12-hour stay—to approximately 80%.¹⁶ The “settling curve” is shown in Fig. 18.

Thus, sedimentation proceeds most vigorously during the first hours of the water’s stay in the basin, and a disproportionately large increase in the capacity of settling tanks is required for any further increase in the amount of matter that settles out.

Water containing earthy or clayey particles of turbidity often requires, for clarification, settling for several days. The processes that take place in this case are more complex, and in each individual case special experiments are required. Colloidal substances settle with particular difficulty.

Prolonged experiments with the sedimentation of drinking water were carried out by Lindley in Frankfurt am Main,³ and led him to the following conclusions:

1) In wide basins the deposits are distributed irregularly and occupy only a part of the full width of the basin.

2) With very slow movement of the water, very small forces are sufficient to produce local currents between the point of entry of the water being clarified and the point of discharge of the clarified water.

3) Water flows along the shortest path; therefore, in wide basins the water at the side walls remains without motion (stagnates).

4) In summer, when the newly entering water is warmer than the water in the settling tank, it tends to flow over the surface of the basin, and the water near the bottom of the basin is forced to stagnate; this phenomenon is observed with the slightest difference in temperature.

Fig. 17.

Fig. 18.

Fig. 19.

On the basis of these experiments by Lindley, settling basins were constructed in Frankfurt am Main for water from the River Main, intended for industrial use and for street sprinkling. These basins consist of separate galleries, 5–6 meters wide, 2–3 meters deep, and 80–120 meters long. During a 6-hour passage of the water through the basin at a velocity of 4 millimeters per second, these settling basins remove from the water about 90% of the suspended matter.

At the present time, the usual residence time assigned for water passing through a settling tank is from 6 to 36 hours⁴, at a flow velocity of 2–4 mm per second, depending on the degree of turbidity of the water. In each individual case, before constructing settling tanks, it is necessary to carry out experiments with the water to be clarified and to determine the most advantageous velocity for it and the time of stay in the settling tank.

From these data the required length \(L\) of the path traversed by the water in the settling tank is determined:

\[ L = 3600\,tv, \]

where \(t\) is the number of hours of residence of the water in the settling tank;
\(v\) is the velocity of flow in meters per second;
\(L\) is expressed in meters.

Lindley recommended giving \(L\) a value of \(260'—400'\).

In designing settling reservoirs it should also be kept in mind that the amount of deposited silt increases as the depth of the water layer decreases, i.e., as the path which the suspended particles must travel in settling to the bottom decreases. But, on the other hand, when the depth of water in a settling tank is excessively small, one may fear its warming in summer and cooling in winter, especially in open settling tanks, which is undesirable.

Therefore the depth of water in settling tanks usually ranges between \(10'—15'\) (3–4.5 meters)¹⁷.

Knowing the daily quantity of water consumed, \(Q\) cubic meters, we find the total capacity of all settling reservoirs, \(V\) cubic meters, which must be sufficient to pass the full daily flow of water at the selected duration of its stay in the settling tank.

Denoting by \(B\) the total width of all compartments of the settling tank and by \(h\) the depth of water in the settling tank (in meters), we find:

\[ V = BhL = \frac{1}{n}\frac{Q}{24}t, \]

where \(n\) is a coefficient expressing the effect of the reduction in the capacity of the settling tank by the sediments deposited in it; \(n = 0.50\text{--}0.75\). Having specified the depth \(h\), this formula is used to determine \(B\).

The calculated total capacity of the settling tank is divided among several of its compartments, each having length \(L\); moreover, the width of each compartment, according to the experiments of Lindley cited above, should not exceed a certain definite value; according to Lindley, \(b = 16'\text{--}30'\).

The number of compartments of the settling tank thus determined is increased (by 15–30%), these additional reservoirs being intended for operation during the cleaning of the compartments of the settling tank (which are cleaned in turn).

In some installations, if it is necessary to settle very fine turbidity, the length of the settling reservoirs, as well as the time the water remains in them (i.e. their volume), is considerably increased. Usually, however, it is more advantageous to remove fine turbidity by filtering the water than by settling. In preliminary settling of water before its filtration (to facilitate the operation of the filters), when very fine turbidity is present in the water, it is likewise more advantageous to add chemical reagents (“coagulants”) to the water before the settling tank than to increase the duration of settling (and, consequently, the capacity of the settling tanks).

The volume of the settling reservoirs is increased in those cases when they also serve as reserve reservoirs; then their capacity, determined on the basis of the requirements imposed on reserve reservoirs, is usually greater than is necessary for settling.

The inlet of water into the settling reservoir, as well as the outlet of water from it, should not be arranged directly through pipes: with a pipe diameter small in comparison with the width of the reservoir, the water will not move uniformly over the entire area of the live section of the settling tank, but will flow (at a speed greater than desirable) in a comparatively narrow stream along a straight line from the inlet pipe to the outlet pipe, while at the side walls of the settling tank immobile masses of water will remain.

In view of this, water from the pipes leading to the settling tank is often first admitted into a gallery whose length is equal to the width of the settling tank (see the diagram in Fig. 19), and from it overflows over the partition separating the gallery from the settling tank, and thus enters the settling tank in an even layer over its entire width. The discharge of water is arranged in a similar manner.

The partitions separating the distributing and collecting galleries from the settling tank are sometimes replaced by shields that move in vertical grooves, which makes it possible to change the position of the water-inlet and water-outlet openings, placing them, as desired, higher or lower than the shield.

Each settling reservoir must be provided with a drain pipe for discharging all the water from it when cleaning the reservoir.

Turning to the consideration of the arrangement of various settling reservoirs, we shall point out that the simplest arrangement is that of each compartment of the reservoir in the form of a rectangle, along which the water flows from one end to the other.

A good example of settling basins of the most widespread type is provided by the basins of the city of Warsaw¹⁷ (see Figs. 20–22).

Each settling basin consists of eight separate galleries, 100 m long and 5 meters wide. The galleries are covered with cylindrical vaults (radius of curvature 3 m). The bottom of the galleries, constructed in the form of inverted vaults, has a slope in the direction of the flow of

\[ \frac{1}{263}. \]

The water in the basin is maintained at a constant level; the depth of the water at the entrance is 2.84 m, and at the outlet—3.22 m.

At the upper end of the basin, along its axis of symmetry, there adjoins a chamber having a width of 3.10 m and a length of 10.25 m, with a depth of 2.70 m. The chamber is divided by a horizontal flooring (platform) into two parts, of which the upper serves as the entrance to the basin, while the lower serves for retaining coarser sediments (sand, etc.), which can be removed by means of hand scoops without interrupting the inflow of water and the operation of the basin. The mouths of the pipelines feeding the basin with water are placed in the lower chamber.

Water is supplied to the basins through branches ($d = 500$ mm) from the main pipeline of raw (i.e., not yet subjected to any purification) water; the branch passes into the chamber, where it ends in two vertical mouths, each 500 mm in diameter, the upper edge of each being 15 cm above the water level in the basin. By passing from one pipe of 500 mm diameter to two mouths of the same diameter, a calm inflow of water into the basin is achieved, without causing agitation of the water in the chamber.

To regulate the inflow of water into the basin, a sluice gate is installed on the branch leading to it.

Besides the two above-mentioned mouths, there is also a third branch in the chamber; the edge of the open end of this branch, introduced into the basin itself, lies 1.85 m below the water level in the basin. This mouth is closed by a special sluice gate, located in the chamber, and serves, when necessary, for flushing the main conduits for raw water by a reverse jet of water from the settling basins, discharged into the Vistula River.

From the distribution chamber the water enters the two middle galleries of the settling basin, and from there is distributed to all the remaining compartments of the basin through openings made in the partition walls at the upper end of the basin; these openings are elliptical in shape, 1.97 m high and 1.18 m wide.

Along the rest of their length the galleries no longer communicate with one another, and the walls separating them adjoin, at the lower end of the basin, the collecting channel.

At the outlet end of the basin, small openings are arranged below in the partition walls for draining all the water from the basin into the waste channel (when cleaning the basin), but during operation of the basin these openings are closed by sluice gates.

The settled water from the galleries overflows into the collecting channel through rectangular openings in the wall. These openings, 50 mm high and 250 mm long, are arranged so that their upper edges lie 100 mm below the highest water level in the basin, and the lower ones 150 mm above the level in the collecting channel. There are 2 such openings in each gallery, i.e. 16 in the basin; they are closed by horizontally moving shields, allowing the size of the opening to be regulated. All 16 sluice gates of one basin are connected with one another by means of a rack with a worm thread on one horizontal shaft. When the shaft is turned, the size of the openings is regulated simultaneously in all 8 galleries of one basin, and thus, at an equal water level in all the galleries (which constitute communicating vessels), uniformity of the outflow of water from them is achieved.

The bottom of the collecting channel has a slope (1/450) in the direction of the pipe (\(d = 0.76\) m), through which the water flows by gravity to the filters.

The channel located below the collecting channel serves to connect the separate basins with one another; before cleaning any basin, it is connected, by means of this equalizing channel, with another basin that has just been cleaned, and the water in them is brought to equal levels: in this way a portion of the water is saved which otherwise would have had to be discharged into the waste channel.

Fig. 20–22 (to p. 47).

In designing the settling basins of the city of Warsaw, it was assumed that the difference in temperature between the water flowing into the basin and the water already in it might cause an uneven distribution of velocities at different depths: it was feared that in summer the newly entering water, being warmer, would flow only along the surface of the basin, while in winter it would tend toward the bottom. In view of this, at the outlet end of each gallery a wall was built, provided at the bottom with openings and at the top with shields; the upper edge of the wall lies 300 mm below the water level in the basin. When the shields are lowered (in winter), the water finds its way to the outlet openings over the wall; when the shields are raised (in summer), it must pass through the openings at the bottom of the wall.

Experience with this device over the course of several years has shown that it is unnecessary, since in Warsaw the difference in temperature between the water of the upper and lower layers is so insignificant that it has no noticeable effect on the uniformity of velocities at different depths. Experience has established that the best results are obtained both in winter and in summer with the shields raised, when the water at the outlet opening flows out from the lower layers.

The velocity of water flow in the settling basin is, on average, 1.45 millimeters per second, and the time the water remains in the settling basin is about 20 hours. Experience has shown that for the water of the Vistula River a velocity of 2 mm is entirely sufficient, at which the time of residence in the settling basin is 12 hours. The basins retain from the water about 76% of the suspended matter. The amount of suspended matter in the water of the Vistula River is, on average, about 90 mg per liter.

The emptying and cleaning of each basin is carried out approximately once a year and lasts 5 days. The cleaning of the distribution chambers by dredges, without stopping the operation of the basin, is carried out every 2 months.

The cost of constructing the settling basins, with a capacity of about 12,000 cubic meters each, ranged from 35 to 25 rubles per 1 cubic meter of capacity (or from 13.5 rubles to 9.6 rubles per 1 square meter of surface); the later structures were cheaper than the first in time of construction, which is explained both by fluctuations in the prices of materials and by the greater experience of the workers and technical staff.

As an example of another design of settling basins, let us mention the “Sachsenhäuser-Berg” reservoir in Frankfurt am Main (Figs. 23 and 24), where, in order to increase the length of the path of the water in the reservoir, a series of partitions has been arranged in plan in a zigzag pattern. The details of the reservoir’s design are clear from the drawing. The total volume of the reservoir is 29,948 cubic meters (2 outer compartments of 7,686 cu. m each and 2 inner compartments of 7,288 cu. m each). The reservoir serves not only as a settling basin, but chiefly as a storage reservoir, which also explains its considerable volume.

Fig. 23.

Fig. 24.

Untreated water.

Clean water.

Fig. 26.

Fig. 27.

Section along \(AB\)

Inlet

Plan

Scale: 10 ft, 0, 50, 100 ft.

Fig. 25.

Left graph: “Number of algae”; horizontal axis: “hours.”

Right graph: “Number of bacteria in 1 cubic cm”; horizontal axis: “hours.”

Figs. 28 and 29.

In Paris, at the waterworks of the “General Water-Supply Company” (“C° Générale des Eaux”), there are settling reservoirs of an even more complex design. The water first passes alongside a series of guide partitions arranged in plan in a zigzag; in this process the heavier suspended particles fall downward. Beyond these partitions there follows a series of successive wide compartments (basins), provided with partitions that block the entire width of the basin, leaving a passage for the water only over the partition or (at the next partition) under it: in this way the water here moves successively now upward, now downward, and, of course, a gradual settling of the turbidity also takes place. Finally, the last compartment of the settling tank serves as an overflow basin. The overall result of the slow, regulated, and winding movements of the water is a more rapid settling of the suspended particles than in quiet sedimentation of water; therefore a smaller area and a smaller volume of settling basins are sufficient.

In America, three or four successive compartments of a settling tank are usually arranged side by side in the form of a rectangle or crosswise (in plan). From the compartment containing dirty water, the stream passes over the dividing wall into the next compartment, and so on.

The design of the covered reservoir in Sundridge Park deserves attention; it consists of concentric compartments (Fig. 25). From the plan it is evident that the water circulates first in one direction, then—in the next circle—in the opposite direction, and so on. The bottom has a slight slope from the circumference toward the center, where the outlet opening for the clarified water is located. Both this reservoir and another of the same type at Eltham and Knockholt work excellently, which confirms the correctness of the principle applied.

We shall also point out that, instead of constructing an artificial settling reservoir, an old riverbed (as in Breslau) or a bend of the river, and the like, is sometimes used, where the velocity of the flow of the water being passed through is brought close to zero. Artificial earthen settling basins with slopes are also sometimes constructed; the bottom and sides, in order to achieve watertightness, must be lined with a layer of clay or concrete; the slopes are sodded or paved. Such reservoirs are usually made open.

Sometimes stone reservoirs are also made open (though rarely, with stone retaining walls and an earthen bottom). Open settling tanks are cheaper than closed ones, and the water in them is exposed to the action of light, which to some extent contributes to its disinfection. However, water in open reservoirs warms up, algae develop in it, dust, insect larvae, etc., get into it; finally, in winter the water in them becomes covered with a layer of ice, which under severe ...

In a humid climate it may attain a considerable thickness, reducing the useful volume of the settling tank. Therefore the construction of covered reservoirs deserves preference.

The best material for constructing covered reservoirs is concrete; brick and stone masonry are also used. In more recent times reinforced-concrete reservoirs have often been made.

The foundation under reservoirs must be continuous, and directly upon it are set the columns that support the vaulted roof of the reservoir.

The vaults are covered from above with a layer of earth 2–3 feet thick, to protect the water from changes in temperature.

In addition to settling basins of the types described, in installations for purifying a small daily quantity of water there are used tanks (usually metal ones) with a whole series of metal partitions in them, which mechanically accelerate the precipitation of sediments from the water (see Figs. 26, 84). The partitions are arranged in such a way as, on the one hand, to lengthen the path of the water in the reservoir, and on the other—to divide the flowing water into layers of smaller depth, since the precipitation of sediments is intensified as the depth of the water flow decreases (i.e., as the distance that a suspended particle must travel to the bottom of the vessel decreases). Finally, the partitions are ordinarily set at an incline, so that the substances settling upon them slide downward to the place where they are removed from the settling tank.

In settling in mechanical settling tanks, the water is passed through them from bottom to top, so that the weight of the sediment particles acts in the direction opposite to that of the motion of the water and facilitates their precipitation.

Mechanical settling tanks are especially often used in the chemical treatment of water, in which a fairly short period (2–3 hours) of settling is usually sufficient, and we shall examine some of their types when becoming acquainted with the methods of softening water.

From the hygienic point of view, settling of surface waters without their further purification can be limited, for drinking water, only to exceptional cases, although when water is settled, together with the removal of suspended impurities, bacteria also partly settle out of it. The example of the Hamburg cholera epidemic of 1892 showed how dangerous it is to rely on the action of settling tanks for drinking water.

The settling of water in sedimentation reservoirs is very useful and important as a preliminary clarification of water before its further purification on filters or by other methods. The costs of constructing settling tanks are more than repaid by savings in the construction and especially the operation of filters, since the preliminary removal of part of the suspended substances makes it possible to increase the filtration rate—

(i.e., to reduce the area of filters required for purifying a given quantity of water per day), and reduces the clogging (silting) of the filters, thus making it possible to clean the filters less often.

When sedimentation is used for the purpose of preliminary clarification of water, there is no need to require from the settling tanks the greatest degree of water purification that settling tanks alone are capable of providing, since for this it would be necessary to increase their volume excessively, far out of proportion to the increase in the degree of purification; the fine turbidity that remains in small quantity (about 10–20% of the total amount of turbidity present in the water before sedimentation) is more advantageously removed by purification on filters or on so-called “prefilters” (preliminary filters of coarse sand or gravel), with which we shall become acquainted below.

Let us note in conclusion that fine clay particles, commonly found in river and lake water during floods (often in the colloidal state), settle with particular difficulty. To remove them one must resort to chemical clarification of the water, adding chemical reagents to the water that form flocculent compounds in the water and, upon settling, carry the clay particles down with them (“coagulation”).

Chapter III.

Filtration of Water.

By filtration of water is meant its purification by passing it through various porous substances.

In nature, natural filtration is very widespread: water, seeping through the earth, is freed from suspended particles, and at the same time from microorganisms; thanks to this, underground waters (groundwaters and spring waters) are, generally speaking, incomparably cleaner than surface waters. In this process, the quality of purification of the percolating water, and the quantity of water that can be purified by a given volume of soil, depend on the properties of the soil and on the rate of percolation. The average rate of natural percolation of water through the earth in a vertical direction is equal to 0.3–0.5 centimeter per day¹. With good fine-grained sandy soil, waters obtained at a depth of 4 meters or more below the surface of the earth usually contain almost no bacteria².

The first artificial sand filters for the purification of water for municipal water supply were constructed in 1829 in London by James Simpson. Thus, this method of water purification, like many other sanitary measures, originated in England. For this reason, sand filters for purifying water passed through them at a low rate are called “English filters.”

These filters are still successfully used, in their essential features, in the very form in which they were invented, and they continue to be improved in various details.

Alongside them, a whole series of other methods of filtration has become widespread, derived from the English method: double or multiple filtration of water through several successive filters; filtration with the preliminary addition of chemical reagents to the water and with a considerable rate of percolation through sand (“American filters”); and, finally, filtration through artificial porous plates of various composition. All these methods, in essence, represent only modifications of the English method of filtration, which is the progenitor of artificial water filtration.

English Sand Filters

General description of the arrangement.

English filters are basins (see Figs. 34 and 35) with watertight walls and bottom (usually of stone or concrete), closed or open, filled with layers of gravel and sand of various grain sizes, the finest sand being placed on top. The water, entering the basin above the sand, seeps through it and is collected by drainage pipes (or channels) laid on the bottom of the basin, beneath the layer of gravel.

The composition of the filtering layers varies somewhat in different installations. The depth of the layer of water above the surface of the sand also varies from 0.50 to 1.30 meters (the desirable depth is 0.90–1.0 m).

When seeping through the sand at a sufficiently low velocity, the water leaves in it suspended matter, as well as bacteria, and enters the drainage pipes already purified.

At the beginning of the operation of each filter (when the filter has been newly built, or has only just been washed), the water is purified by it very unsatisfactorily; as the filter continues to operate, the quality of the filtrate (i.e. of the filtered water) improves, and after some time of operation of the filter the water purified by it becomes suitable for water supply; they say that the filter has “ripened.”

English filters can retain from the filtered water up to 90–98% of the suspended impurities and bacteria contained in it.

After the filter has operated for some time (about 2–6 weeks), the filtering material (sand) becomes clogged and ceases to purify the water well; at the same time, the permeability of the filter decreases, and in order to pass water through it at the same rate as at the beginning of the filter’s operation, a considerable head of water is required (i.e. a difference between the water levels above the sand surface and in the clean-water collecting chamber receiving the water from the filter drainage).

The filter has become fouled and requires cleaning: the operation of the filter must be stopped, the water drained off, and the upper layer of sand, which is especially heavily fouled (1–3 cm), removed.

Then a new period of the filter’s operation begins; it must ripen again (and during ripening the water should be discharged to waste, as being unsatisfactorily purified), and so on.

When, after several cleanings, the thickness of the sand layer has decreased to a certain limit, it is replenished with new, clean sand.

Theory of the Action of Sand Filters

The basis for explaining the action of sand filters was laid by the investigations of Piefke (around 1886), who applied the methods of bacte-

biological analysis to the systematic study of the process of filtration¹⁵.

The interest of bacteriologists in the processes taking place in English filters is entirely understandable, since the result of these processes is a considerable reduction in the number of bacteria in the filtered water, amounting to 90–98% of their initial quantity.

The retention of bacteria can in no case be caused by the physical properties of the sand itself¹⁶.

Grains of sand adjoining one another, even in very fine and dense sand, cannot by themselves retain bacteria from the water passed through the sand, since the intervals between the grains of sand are always exceedingly large in comparison with the size of bacteria; this is easy to see by examining sand under a microscope. If, for example, grains of sand no larger than 0.3 mm³ are moistened with water and a pure culture of Bacillus megatherium, which are known as one of the largest bacteria (length up to 5 microns, i.e., thousandths of a millimeter), is mixed with them, then under the microscope we obtain the image schematically represented in Fig. 27; the relative dimensions are fully preserved in the diagram. From the drawing it is clear how small, in comparison with the grains of sand, are the rods of B. megatherium, which can easily slip between grains of sand that are not always in complete contact with one another.

All investigations of submerged English filters have shown that at the beginning of operation the filter does not retain microscopic germs at all, and retains only a small part of the larger organisms and suspended matter. With further operation of the filter, the ratio of the amount of suspended matter passing (slipping) through the filter to the amount retained gradually decreases until, after a certain interval of time, it becomes negligibly small. Fig. 28 gives a diagram of the suspended matter passed by a filter, for a filter purifying river water. The decrease in the number of bacteria in the filtered water during the first hours of operation of the filter is shown in Fig. 29.

The gradual improvement of the filtered water is explained by the accumulation on the surface of the sand of a deposit formed from settled and retained substances (organic residues, silt, etc.), which coats the grains of the upper layer and closes the intermediate channels, forming a soft partition (film), not strongly stretched, but very elastic.

On the forming film there develop numerous microscopic species—algae, colonies of bacteria, and so forth. The surface—

the sand forms a continuous, more or less permeable fabric of plant fibers, for the most part lying on the very surface and penetrating into the layer of sand only to a depth of a few centimeters. It is precisely this “filtering film” that constitutes the part of the filter which actually possesses filtering capacity, especially with respect to the retention of bacteria.

Only after the loose fabric of the film has reached a certain consistency does the water, passing through its pores, become freed from suspended matter and even, to a considerable degree, from bacteria; and the filter may be regarded as “ripened,” and will work satisfactorily so long as the film remains continuous and undamaged.

Numerous investigations have established that the formation, on English filters, of a filtering film capable of retaining bacteria takes place with the participation of biological processes (i.e., is the result of the vital activity of microorganisms). As for the process by which bacteria are retained from the filtered water by the ready-formed, already ripened film, this process may at present be regarded as occurring mainly by mechanical means, although, undoubtedly, there is also a biological action of the ready-formed film on the substances present in the filtered water and retained from it.

The significance of algae for the operation of the filter was studied (in Hamburg) by Strohmeyer⁴, who, over the course of an entire year, examined the films of filters being cleaned and determined the species of algae (algae); there proved to be more than 160 of them. Strohmeyer demonstrated the importance of algae (especially of the family Bacillariacea) for filling the gaps between grains of sand and for accelerating the formation of the film. Further, he demonstrated that green algae (containing chlorophyll), as they develop, vigorously destroy bacteria and hinder their multiplication, acting on them, in all probability, by the oxygen generated by the algae in the nascent state, which also exerts a strong oxidizing effect on organic substances.

Two of the most widespread species of algae are shown in Fig. 4.

Algae develop more vigorously under the action of light; consequently, open filters ripen more quickly, but also become fouled more quickly than closed ones.

Further, the biology of sand filters was studied by Dr. Kemna in Antwerp¹⁷. Like Strohmeyer, he noted the appearance of different species of algae at different times of the year. He also pointed out that floating algae on the surface of the water on a filter usually serve as a sign—

the sign of lagging behind in some places of the filtering film relative to the sand, i.e., ruptures of the film, through which bacteria pass through the filter.

According to Kemna’s expression, the loose layer formed by sprouting algae and by their interlacing fibers catches bacteria like an insect’s web.

Finally, according to Kemna’s experiments, one of the means by which bacteria are retained is the sticky mud film, consisting partly of dead and partly of living material; particles of dirt, as well as germs, adhere to the grains of sand.

Thus, we see three modes of the filtering action of the film: retention of bacteria by a crust of interlacing algae and fibers; adhesion of bacteria to the sticky mud film that has grown on the sand⁵; and, finally, oxidation of bacteria by oxygen released by the algae.

The role of the sand. The filtering film is nevertheless not the only filtering element of sand filters; this is evident from the fact that the results of filtration improve with an increase in the thickness of the sand layer, as was shown, for example, by the experiments in Lawrence (America), in which clean water was obtained by filtration through a layer of sand 1.20 meters thick, with a grain size of 0.15 mm, in the absence of a filtering film on the sand.

Piefke also showed that bacteria are retained not only by the film, but also by the entire thickness of the filtering layer. In Fig. 30 the lengths of the black horizontal lines represent the relative quantities of bacteria remaining in the filter sand at different depths.

It should be noted that, in order to determine the number of bacteria retained by individual layers of sand from the water being filtered, experiments must be carried out with specific bacteria (Piefke used cultures of Bacillus violaceus; in Lawrence they used B. prodigiosus), since bacteria belonging to the species that normally inhabit water may be retained from the filtered water by the upper layers of sand or by the film and then again enter the filtrate from the lower layers of the filter: bacteria can multiply in the sand of the filter itself, having arrived there from previously filtered water, and from there penetrate into new portions of the water being filtered, which constantly occurs⁶.

Let us note that, of all the layers of sand and gravel of various coarseness that fill the filtering basin, only the upper, fine-grained layer of sand plays a role in the filtration process and is called the “filtering layer”; the other layers of coarser sand and gravel are laid merely in order to support the upper layer and to prevent grains of sand from being washed by the filtered water into the drainage channels; these layers are called “supporting” layers.

Filtration experiments through sterilized sand, carried out by Piefke, showed that the filtration process proceeds worse than with non-sterilized sand, and that the “ripening” of the filter is substantially necessary. However, the majority of investigators at present hold the opinion that, although microorganisms do participate in the filtration process, the chief role is nevertheless played by the mechanical retention of bacteria (and also, of course, of other suspended substances), and that the “ripening” of the filter amounts, in essence, to compaction of the sand, filling of the spaces between the grains of sand, and formation of a film. Thus the “ripening” of a sand filter cannot be equated with the ripening of biological filters—oxidizers for the purification of wastewater, which consists in the multiplication of aerobic microorganisms that oxidize and mineralize the products of decomposition⁷. Finally, the role of the film is also reduced chiefly to mechanical effects on bacteria (with the exception of their oxidation by algae).

All that has been said above makes it possible, in the process of filtration, to regard bacteria as one of the kinds of suspended matter (though, to be sure, possessing certain special properties), and not as living organisms, in contrast to the processes of sterilization and disinfection of water, in which the aim is to kill the bacteria, not to catch them.

The mechanical theory of the action of filters. On the basis of prolonged observations of the operation of the filters of the city of Warsaw, engineer L. K. Baginsky¹⁸ proposed a mechanical theory of the action of sand filters, pertaining to their retention of all sorts of suspended substances (including bacteria). This theory is very interesting as one of the attempts to bring the complex process of filtration under mathematical formulas.

L. K. Baginsky regards both the sand and the film as consisting of a series of horizontal layers differing in coarseness

particles; between the particles of sand (or film) there are capillary channels, having a series of contractions and expansions. When the filtered water passes through the capillary channels, their contractions and expansions cause changes in the velocity of the water’s passage and a loss of the water’s living force, which, in turn, entails the settling of the substances suspended in the water. The smaller the particles of the sand layer, the narrower the capillary channels between them, and the better the layer retains the impurities carried by the water. From the point of view of this theory, the film differs from the filtering sand only in the size of the particles and of the channels between them.

If \(x\) is the total content of suspended impurities in the water entering the filter, then a certain fraction of the impurities, \(ax\), is retained on the surface of the film, and there penetrates into the thickness of the film a content of impurities:

\[ y = x(1-a). \]

With the thickness of the film \(\delta_1\) and the average dimension of the smallest particles composing it \(d_1\), and with the total thickness of the filtering layer of sand \(\delta_2\) and the average size of its grains \(d_2\), the water passes through the thickness of the film in \(\frac{\delta_1}{d_1}\) layers, and through the thickness of the sand in \(\frac{\delta_2}{d_2}\) layers; and in each layer the water will leave some fraction \(a_1\) of those impurities that were contained in it on entering that layer. In this case the relative quantity \(a_1\) of impurities retained by each layer of the film is greater than the relative quantity \(a_2\) of impurities retained by each layer of sand (because the particles of the film are smaller than the grains of sand).

Thus, in the first layer of the film \(a_1 y\) impurities will be retained, and into its second layer there will pass:

\[ y - a_1 y = y(1-a_1)=y_1 \]

Similarly, in the second layer \(a_1 y_1\) will be retained, and into the third layer there will enter

\[ y_1(1-a_1)=y(1-a_1)^2 \]

After passing through \(\frac{\delta_1}{d_1}\) layers of the film, there will enter the upper layer of the filtering sand

\[ y_2 = y(1-a_1)^{\frac{\delta_1}{d_1}} \]

impurities.

Analogously, after passing through the first layer of sand the water will contain \(y_2(1-a_2)\) impurities, and after passing through its entire thickness, i.e. all its \(\frac{\delta_2}{d_2}\) layers, there will remain in the water impurities:

\[ z = y_2(1-a_2)^{\frac{\delta_2}{d_2}} = y(1-a_1)^{\frac{\delta_1}{d_1}}(1-a_2)^{\frac{\delta_2}{d_2}} = x(1-a)(1-a_1)^{\frac{\delta_1}{d_1}}(1-a_2)^{\frac{\delta_2}{d_2}}. \]

This formula indicates that the quantity of impurities remaining in the filtered water is the smaller (i.e., the better the purification), the greater the degree indices \(\frac{\delta_2}{d_2}\) and \(\frac{\delta_1}{d_1}\), i.e., the finer the sand and the thicker the filtering layer of sand and film. These conclusions are fully consistent with practice.

Thus, according to this theory, the filtering film is regarded as a layer of fine-grained material (silt), entirely without reference to its specific properties; it is clear that these properties are not amenable to mathematical accounting and cannot be expressed by a formula.

Concluding the consideration of the theory of the action of filters, let us note that even dissolved substances, especially colloidal ones, are partly removed from the water as it passes through English filters⁸. The appreciable reduction in the content of organic impurities achieved by some filters in the purification of river water can also, with great probability, be ascribed to the removal of dissolved colloids.

The mud coating of the grains of the filter has the ability to absorb (draw into itself) dissolved colloids and to retain part of them. In the opinion of the investigators Kamman and Cornwath, the filtering film does not precipitate colloids from solution, but causes biological changes in them (fermentation, oxidation, decomposition), the consequence of which is the subsequent removal of colloids.

The change in the chemical composition of water during filtration is expressed in a reduction in the content of organic substances, ammonia, and iron, and in a certain increase in nitrates. This is evident both from the diagram of the operation of the Zurich filters (Fig. 58) and from the following data obtained in experiments at Lawrence.

NAME.	Content in water, mg per liter: Before filtration	Content in water, mg per liter: After filtration	Decrease, %
Organic substances (by oxygen absorbed)	3.0	2.8	28.2
Free ammonia	0.084	0.068	19.0
Albuminoid ammonia	0.202	0.109	46.0
Nitrates (by nitrogen content)	0.14	0.31	—
Nitrites (by nitrogen content)	0.003	0.005	—
Number of bacteria in 1 cubic cm	14000	258	98.16

Thus, English filters exert an oxidizing action on the water being filtered. In other respects they do not change its chemical composition.

Fig. 31.

Fig. 30.

Fig. 32.

Fig. 33.

Influence of the Thickness of the Filtering Layer, the Grain Size of the Sand, and the Rate of Filtration on the Operation of the Filter.

The thickness of the filtering layer, as well as the thickness and composition of the supporting layers, vary greatly in English filters in different cities, as may be seen from Fig. 31, which shows in section a series of filters. In England the filtering layer is made about 1 m thick; in Germany it is often much thinner. However, a thickness of 0.80–1 m should not be considered excessive, since it is desirable to be able to clean the filter several times without adding new sand. At each cleaning it is necessary to remove the contaminated upper layer of sand, 10–20 mm thick; the interval between cleanings varies from 2 to 6 weeks, depending on the quality of the water being filtered, and on average may be taken as 3 weeks.

The smallest permissible thickness of the filtering layer also depends on the properties of the water; generally speaking, it is undesirable to take it as less than 0.40–0.50 m. The initial thickness of the layer should be set so that new sand need be added no more often than once a year.

At present there is a tendency, especially in America (after the experiments at Lawrence), to bring the thickness of the filtering layer of fine sand up to 4′ (= 1.22 m), while at the same time reducing the thickness of the supporting layers, which, instead of 0.50–0.70 m in the former filters, is brought down to 1′ = 0.30 m, or is even replaced directly by porous supporting slabs of lean concrete 0.07 m thick, as was done in Paris (Fig. 32).

The sand constituting the filtering layer should be as clean as possible; it is often washed beforehand. The generally accepted grain size ranges from 0.3 to 1 mm. In this connection, the grain size (or, as it is called, the “effective size”) of the sand is taken to be⁹ the size of the openings of such a sieve through which not more than 10% of the volume of the given sand passes.

Sand for a filter should be of uniform grain size (homogeneous).

In view of the fact that the degree of uniformity plays an important role in filtration, one should always take into account the “coefficient of uniformity” of the sand. By the coefficient of uniformity is meant (according to the proposal of Allen Hazen, superintendent of the filters in Albany) the quotient obtained by dividing the diameter of the largest grain among the coarsest 60% of the smallest particles by the diameter of the largest grain among the coarsest 10% of the smallest particles.

In America it is not customary to use sand whose grain size (according to the above definition) exceeds 0.34 mm, and whose coefficient of uniformity would be less than 1.7 or greater than 3.0.

As an example, let us cite the requirement adopted for the filters of the city of Washington: “The grains of sand must, in their composition, be of hard rock, incapable of dissolving, and their diameters must be as follows: no more than 0.5% and 1% by weight of these grains may have a diameter smaller than 0.13 mm; no more than 8% by weight may consist of grains smaller than 0.26 mm; at least 70% by weight must have a diameter smaller than 0.83 mm, and no more than 90% a diameter smaller than 2.1 mm. No particle may have a diameter greater than 5 mm.”

The quantity of water that seeps through a layer of sand per unit time naturally depends on the grain size of the sand, on the thickness of the layer, on the head during percolation (i.e., on the difference between the water levels above the sand and in that compartment of the filter into which the purified water enters through the drainage channels), and on the temperature.

To express the rate of percolation, the following formula has been proposed¹⁹:

\[ V = c d^{2} \frac{h}{l} (0.7 + 0.03 t) \]

where \(V\)—the filtration rate (in other words, the height of the layer of water passed through the filter), in meters per 24 hours,

\(d\)—the mean diameter of the grains of sand, in millimeters,

\(h\)—the loss of head,

\(l\)—the thickness of the filtering sand layer, in the same units as the loss of head \(h\),

\(t\)—temperature in degrees Celsius,

\(c\)—a coefficient close to 1000.

This formula indicates the important influence of temperature on the productivity of a filter: all other conditions being equal, at temperatures of \(0^\circ\), \(10^\circ\), \(20^\circ\), and \(30^\circ\), the quantity of water seeping through the filter changes respectively in the ratio: 0.70; 1.0; 1.30; and 1.60.

In practice, the filtration rate is varied from 5 to 30 cm per hour (2″—12″ per hour), depending on the quality of the water being filtered; for water of average quality, the normal rate is considered to be 10 cm (=4″) per hour.

The number of bacteria in the filtered water, for evaluating the operation of the filter, is usually expressed as a percentage of their initial content in the water being purified.

Experiments in Lawrence led to the conclusion that the ratio of the bacterial content in the filtrate to their initial content in the raw

water, expressed as a percentage, depends exclusively on the filter and the rate of filtration and can be represented by the formula:

\[ N\%=\frac{1}{2}\frac{q^2 d}{V l}; \]

where \(d\)—the average diameter of the sand grains,

\(l\)—the thickness of the filtering layer,

\(q\)—the quantity of water passing through the filter per day, in millions of gallons per acre of filtering surface (1 million gallons per acre = 0.93 cubic m per 1 sq. m). In practice, in existing filters the bacterial content in the filtrate is usually greater than that calculated by this formula.

In this formula the percentage ratio \(N\) is taken as independent of the bacterial content in the raw water fed to the filter. Meanwhile, the bacteria found in filtered water belong only in part to the number of bacteria that were in this water before filtration; some, however, enter the water from the filtering sand²⁰; therefore, when water containing very few bacteria is passed through a filter, the number after passing through the filter sand may even increase. Hence it is clear that a direct dependence between the number of bacteria in the water before and after filtration can by no means always exist, and one must agree with the opinion of some investigators²¹, who believe that, all other conditions being equal, the bacterial content of raw water affects their content in the filtrate only in the case of a high initial content of them (over 100,000–200,000 in 1 cu. cm); in the case of a smaller number of bacteria in the raw water, no direct dependence between these quantities exists.

Design and Calculation of English Filters.

Before describing the arrangement of English filters, we shall give the general instructions set forth in the directive of the Imperial German Health Office (Kaiserlicher Gesundheitsamt), mandatory for municipal waterworks²².

1) The construction of sand filters is advisable only if a properly moderate rate of filtration is ensured, corresponding—

...corresponding to the properties of the water being filtered; therefore it is necessary that the area of the filters be fully sufficient.

Each section of the filter must have devices allowing it to be disconnected from the conduit for clean filtered water and the filtrate from it to be discharged into the waste pipes, if this filtrate proves to be of poor quality.

3) Each section of the filter must be arranged so that it can be controlled independently of the other sections, and also so that independent checking is possible of the inflow of water to it, of the head, and of the composition of the water. There must also be devices for the complete emptying of each section independently of the others, and for filling it with water (after each cleaning) from below upward to the surface of the sand.

4) It is necessary that the filtration rate in each section of the filter be capable of being brought to the conditions most favorable for the process of filtration, be as uniform as possible, and be protected from rapid fluctuations. For this purpose the normal fluctuations caused by changes in water consumption during the day should, as far as possible, be evened out by a storage basin (reservoir) ¹⁸).

5) The filters must be so arranged that their operation is not affected by a variable water level in the clean-water reservoir.

6) The head of water on the filter must never be so great that ruptures of the filtering film can occur. The limit to which the head may be raised without harm to the filtrate must be determined for each installation by bacteriological investigation.

7) The filters must be so arranged that all parts of the surface of each filter operate as uniformly as possible.

8) The walls and floor of the filter must be watertight, and in particular the possibility must be excluded of unfiltered water penetrating into the channels for filtered water. For this last purpose it is necessary to construct and maintain as watertight the air shafts of the clean-water channels.

9) The thickness of the sand layer must be sufficient so that under no circumstances is it reduced by cleanings to less than 30 centimeters; moreover it is always recommended, when the operating conditions of the filter permit, not to reach this limit and to maintain a minimum sand-layer thickness of 40 centimeters.

10) For continuous hygienic control of the water supply it is recommended everywhere, where the available forces allow it, to examine daily the filtrate of each section of the filter. This is especially important:

¹⁸) The volume of this reservoir is calculated, like the volume of any storage reservoir, according to fluctuations in water consumption at different hours of the day.

a) after the construction of a new filter, until its operation has become entirely normal;

b) at every filling of the filter with water after its cleaning—at least two days longer than the period needed to obtain a filtrate of satisfactory composition;

c) after the pressure on the filter reaches more than two-thirds of the established maximum pressure;

d) when the pressure above the filter suddenly falls;

e) under all exceptional circumstances, for example during freshets and floods.

11) To make possible the bacteriological investigations indicated in the preceding point, it is necessary that the filtrate from each compartment of the filter be accessible for taking water samples at any time.

When beginning the design of English filters, one first determines their filtering area \(S\). If \(Q\) is the quantity of water to be filtered per day, and \(V\) is the selected filtration rate in meters per hour, then the total working surface of the filters must be equal to

\[ S=\frac{Q}{24V}. \]

Usually \(V\) is taken to be not more than 0.10 m per hour; preliminary filtration experiments with the water for which the filters are being designed are desirable.

The total surface \(S\) is divided among several separate filter basins, the area of each being made about 2000–2500 sq. m for covered filters and 3300–4300 sq. m for open filters¹⁰. Denoting the area of one basin by \(s\), we obtain the number of compartments

\[ N_1=\frac{S}{s}. \]

The resulting number \(N_1\) of compartments must be in operation at every given moment; but since each filter compartment is from time to time subjected to cleaning, it is necessary to construct not \(N_1\) compartments, but more, \(N\). If the filter (gradually, by compartments) is cleaned \(n\) times per year¹¹, and each cleaning takes \(t\)¹² days, then each filter compartment operates only \((365-nt)\) days per year;

i.e. over the course of

\[ \frac{1}{m}=\frac{365-nt}{365} \]

part of a year, and all \(N_1\) compartments together will purify, over the course of a year, only

\[ \frac{1}{m} \]

part of the entire required quantity of water. Therefore, to purify the full annual consumption of water, it is necessary to build not \(N_1\) compartments, but \(m\) times as many; that is, there is required

\[ N=\frac{365}{365-nt}N_1=\frac{365}{365-nt}\cdot\frac{S}{s} \]

compartments, each of area \(s\).

The number \(N\) thus found is usually increased by 1 or 2 compartments, in case repairs are needed.

For small installations, no fewer than 3 or 4 compartments are made.

The filter compartments are usually given, in plan, a rectangular shape, with a ratio of sides from \(2:1\) to \(3:1\).

The walls and bottom of the filter are made of stone or brick masonry on cement, or, most often, of concrete. If the filters are covered, then, to support the covering, concrete (or stone) columns are installed at intervals of 3–4 m from axis to axis and with a cross-section of about 0.30–0.40 m square.

These columns are spanned by arches, likewise brick (half-brick) or concrete; the rise of the arches is about \(1/4\). Sometimes reinforced-concrete coverings are used. Inspection wells and ventilation openings are made in the covering. The covering is filled over with earth on top (in a layer 3–4 feet thick).

The outer walls are often arranged as a row of cylindrical vaults with vertical axes, their concavity facing into the filter; these vaults bear against brick or concrete columns, which take the horizontal load of the walls (the earth pressure). This arrangement achieves a saving in the volume of wall masonry and reduces the formation of small cracks in concrete walls when the temperature changes (as practice has shown).

The bottom of the filter also often has the form of an inverted vault (between the longitudinal rows) and is laid on a layer of kneaded clay or clay concrete. The bottom is given a slope in the direction of the water drain.

The height of a closed filter from the bottom of the basin to the ceiling is made about 3–3.50 meters; in this case the distance from the surface of the sand to the ceiling must be not less than 1.70 m, for convenience of cleaning. The height of open filters (to the top edge of the walls) is made somewhat less (2.10–3 m). The height of the basin is composed of the heights of the drainage channel, the supporting layers (gravel and coarse sand), the filtering layer of sand, the water layer, and the reserve space above it.

Drainage. To collect the filtered water, drainage pipes are laid on the bottom of the filter; most often, along the middle of the filter compartment, parallel to its length, a main outlet channel is arranged, and perpendicular to it, at a distance of 2.5–4 meters from one another, water-collecting drains are laid (see Fig. 33). The main channel is usually made of concrete or brick; sometimes its dimensions vary along its length, as in the filters in Albany (Fig. 33), but more often it is given the same cross-section throughout. A glazed earthenware pipe may also serve to form the channel.

The main channel is sometimes arranged at the lower end of the filter (as in Warsaw, see Figs. 34 and 35), across it, laying the water-collecting drains along the filter compartment and giving the bottom between the drains a slope toward them; in this case the water-collecting drains are made large. In Warsaw the drains are 0.50 m wide and 0.45 m high; their side walls are made of bricks laid on edge; vertical gaps of 5 mm are left between the bricks, through which the water enters the channel. The brick walls are covered with concrete slabs.

There is also a diagonal arrangement of outlet drainage channels, if the filter compartment has the form of a square in plan (Paris, Ivry filtration station; for a section of this filter see Fig. 36).

The collecting drains are drainage pipes 10–15 cm in diameter, or earthenware pipes 10–20 cm in diameter, provided in the upper semicircumference with openings, or, finally, channels made of solid or hollow brick. A very convenient and simple arrangement of collecting drains is the Muir method: on the bottom of the filter, channels are formed by bricks laid in courses, half a brick high, with spaces between them, covered also with bricks laid flat, with open joints (Figs. 36 and 37).

Supporting layers. To facilitate the inflow of water to the collecting drains, and also to prevent the fine sand composing the filtering layer from penetrating into their openings, a layer of gravel 2.5–5 cm in diameter is placed around the drains (above and at their sides), followed by 3 or 4 layers of finer gravel, of gradually decreasing diameter (usually the diameter of the grains of the upper course is made half that of the preceding course). Each subsequent layer must cover the preceding one well; for this a thickness of 4–8 cm is sufficient. These layers are called supporting layers. The size of the grains of the upper supporting layer must be such that the spaces between the grains are smaller than the grains of sand of the filtering layer.

Depth of water above the surface of the filtering layer, as experience has shown, should be approximately 0.90–1.0 m. At such a depth

Fig. 36 and 37.

Labels in the section of the filter bed:

• Water to be filtered
• Sifted Loire sand
• Washed gravel
• Washed ordinary pebbles
• Large washed pebbles laid by hand on two courses of superposed bricks
• Foundation slab

Fig. 35.

Section A–B.

Scale: 0–3 m.

Fig. 39.

the surface of the sand and the filtering film are subjected to the minimal influence of waves and currents that may appear in the water of the filter basin, and it is possible to achieve a fairly high loss of head in the filter without deterioration of the quality of the filtrate.

The thickness of all the indicated layers, as well as the depth of water above them in the various installations, is shown in Fig. 31; we shall again give sections of the filters of Moscow (the Rublev waterworks) (Fig. 38), Warsaw (Fig. 35), and Paris (Figs. 36 and 32).

In Moscow²³, above the drainage channel the following layers are laid:

1) gravel of 20–30 mm size, 7 dm thick.
2) gravel of 10–16 mm size, 7 dm thick.
3) gravel of — 4 mm size, 4 dm thick.
4) sand of — 2 mm size, 2½ dm thick.
5) filtering layer of sand of 0.50 mm size, 3′5″ thick.

The thickness of the layer of water being filtered is 4 feet.

In Warsaw (Figs. 34 and 35), coarse gravel (35 mm in diameter) is laid around the drainage channels, then gravel of medium size (20 mm), then a layer of fine gravel (8 mm), with a total thickness of 0.40 m; above them is a layer of coarse sand 0.10 m thick, and finally a filtering layer of 1.0 m. The greatest height of water above the sand is 0.95 m.

In the drawings 32 and 36 we have sections of the open filters of the city of Paris (Ivry), old and new (1904): with the same basin depth (2.10 m) and depth of the water layer (0.90 m), the old filters have a filtering layer of only 0.50 m, and beneath it (from top to bottom):

a layer of coarse-grained washed sand 0.10 m thick.
a layer of medium gravel of 0.15 m.
a layer of coarse gravel about 0.20 m.
drainage channels of bricks laid flat, 0.06 m high.
(this row of bricks is covered by a second row, between the bricks of which the above-mentioned coarse gravel is poured).

In the new filters the entire supporting layer (which had a thickness of 0.45 m) was replaced by porous slabs of lean concrete, measuring 0.50 × 0.50 × 0.07 m, laid directly on the lower layer of brick; the total height of brick and slabs was 0.13 m, which made it possible to increase the thickness of the filtering layer to 0.87 m.

The porous slabs, as experience has shown²⁴, allow water to pass through quite well, while at the same time retaining fine sand. Into the slabs, during manufacture, rods of round iron are embedded (in number

Fig. 34a.

To p. 70.

Labels in the drawing: \(D0{,}50\); \(D0{,}40\); \(D0{,}40\).

Scale \(1:200\): 0 1 2 3 4 5 6 7 8 9 10 m

Fig. 34 b, c.

Fig. 38a.

Labels visible in the figure:

• Elev. 19.14
• Elev. water level = 12.00
• Elev. 11.43
• 1st layer
• 2nd layer
• 3rd layer
• 4th layer
• 5th layer
• Collector 2

Fig. 38b.

Labels visible in the figure:

• Chamber of the inlet pipe
• Gate of the filter
• [[unclear: handwritten label concerning the drive and the outlet trough of the filter]]
• [[unclear: handwritten label inside tank, possibly “tank of the pipe compartment”]]
• 84.5 cm

Fig. 44.

Labels visible in the figure:

• Gate of the filter
• Cement layer
• Plaster
• Drainage channel
• 1st layer
• 2nd layer
• 3rd layer
• 4th layer
• 5th layer

b), which allows them to bear the load of the layer of sand. Besides reducing the required height of the basin (with a constant thickness of the filtering layer), porous slabs offer still other advantages: the outflow of the filtrate becomes more regular; owing to the uniformity of the sand loading of the filter, inspection of the filter and repair of its walls and bottom are facilitated.

Let us note one more detail in the construction of the new Paris filters: along the perimeter of the basin, at the walls, a draining channel is arranged, carrying water away into the waste pipes: water near the walls of filters usually slips along the wall, bypassing the filtering layer and without having been filtered²⁵; in the absence of the indicated channel, this water enters, through the common drainage, the conduits of clean water.

The loss of head during filtration is composed of the losses in the filtering layer, the supporting layers, and the drains.

The loss of head during percolation through the filtering layer is calculated by the formula

\[ V = c d^{2}\frac{h}{l}(0.7 + 0.03t) \]

which we gave on p. 65.

The loss of head in the supporting layers is extremely small, and it may be neglected. The loss of head in the drainage pipes can be determined by the formula

\[ h = \frac{4}{c^{2}}\cdot\frac{V^{2}}{d}\cdot l \]

If the dimensions are expressed in meters (\(v\)—in meters per second), then the coefficient \(c\) for drainage pipes is calculated as a function of the diameter:

\[ c = \frac{100\sqrt{d}}{0.7 + \sqrt{d}} \]

In Fig. 33, which shows the arrangement of the drainage pipes of the filter in Albany, lines of different losses of head in the drains are depicted for water filtration at a rate of 10 cm per hour; as can be seen, those particles of water which enter the drains at places farthest removed from the outlet opening lose only 6 mm of head in flowing to the outlet from the filter.

In general, the distance between drains and the dimensions of the drains are calculated so that the flow of water through each drain corresponds to a loss of head of no more than 10 mm.

Regardless of considerations of head loss, the distance between drains must not be made excessively large, also because then, owing to the tendency of the particles of the water being filtered to move toward the drains along the shortest path, not the entire mass of the filtering sand would work uniformly, but only funnel-shaped masses of sand above the drains.

It is desirable not to allow the total loss of head on the filter to exceed 90 cm (3 feet), since a greater head can easily cause ruptures of the filtering film; and, in general, with a strong flooding of the filtering layer the quality of the filtrate deteriorates. In America a head of up to 6′ (180 cm) is sometimes permitted, when a thick layer of filtering sand (4′) is used, which could satisfactorily filter the water even in the event of ruptures of the film.

For each filter the greatest permissible loss of head is established by observations of its operation and by bacteriological analyses of the filtrate.

Devices for admitting and discharging the water being filtered. The admission of untreated water onto the filter must be carried out in such a way as not to disturb the water present in it, and through it the filtering film, which, as we know, is easily ruptured; in other words, the velocity of the entering water must be as low as possible. One method for this consists in terminating the pipe supplying water to the filter with two vertical branches (placed within the thickness of the sand), the diameter of each of which is equal to the diameter of the pipe; the upper ends of the branches are widened (their edges are bent outward) (Fig. 39). The velocity of water flow in the feed pipe is made 30 cm per second or less. Sometimes (Fig. 40), on the surface of the sand at the point where the pipe emerges, in order to avoid scouring, an iron sheet is laid, or else a layer of cobblestone.

Ordinarily the pipe delivering water to the filter from the source (or from the settling basins) enters an inlet chamber separated from the filter, in which the water loses its velocity; the pipe supplying water to the filter leaves only from the inlet chamber, i.e., under a slight and only slightly fluctuating head.

One of the types of inlet chamber is shown in Figs. 41 and 42: the chamber consists of an iron cistern divided vertically by a partition into two parts and placed in a concrete well, the height of which is equal to the height of the filtering layer. The height of the cistern is above the water level on the filter. The water enters the first compartment of the cistern, loses its velocity in it, overflows across the partition into the second compartment, and only through an opening at the bottom of its outer wall does it pass out into the concrete well. This well protects the sand from scouring.

To maintain a constant water level on the filter, a valve with a float is installed on the pipe supplying the water (see Fig. 40), which, when the water level on the filter rises, reduces the inlet ...

Fig. 41.

Fig. 43.

Fig. 42.

Visible labels in the diagram: \(a\), \(b\), \(c\).

Fig. 49.

Visible labels in the diagram: Filter; Sluice; Head chamber; water; Font; of the Eure. Lettered designations: A, B, C, E, F, H, M, N, P.

Fig. 48.

Labels in the figure:

• Filter
• Main drain
• Water-level indicator on the filter
• Water-level indicator in the outlet chamber
• Water-level indicator in the gauge chamber
• Outlet chamber
• Gauge chamber
• Overflow
• Departure
• Valve [[unclear: drain valve]]
• Constant water level
• 38.75
• 38.60
• 38.30
• 38.15
• 37.45
• 37.21
• 36.95
• M
• N
• R
• B
• g

Fig. 40.

Labels in the figure:

• Chamber
• Float regulating the openings b and c
• Arrival
• Sand
• Entry of water onto the filters
• Facing
• 38.75
• 38.30
• 35.82
• 35.20
• c
• h

the opening of the pipe, and, when the level is lowered, increases it. However, at many filtration stations the regulation of the water inlet is done by hand.

The devices regulating the inlet are placed in the inlet chamber. During the entire operation of the filter it is desirable automatically to maintain a constant filtration rate, since any change in the filtration rate worsens its results, making them variable, and at the same time makes the output of the filter variable. Therefore it is necessary to arrange an appliance for the inlet and outlet of the water being filtered in such a way that the filtration rate remains unchanged, and changes in it could be made, if necessary¹³, only by hand.

The rate of percolation of water through the filtering layer, as we know, depends, for each given filter, first, on the thickness of the film and the degree of silting of the sand and, secondly, on the height of the head. During operation of the filter the film thickens, and the percolation of water becomes increasingly difficult; therefore, in order to keep the rate constant, it is necessary to increase the head of water, i.e., the difference between the water levels above the filtering layer and in the outlet chamber of the filter (the outlet chamber is the compartment into which the filtered water enters from the drainage channels, and from which it is conveyed through pipes to the clear-water storage reservoir).

Thus, in order to keep the filtration rate constant, it is necessary either to raise the level of the water above the sand, or to lower the level in the outlet chamber, or, finally, to do both simultaneously.

In the first case the outlet chamber is divided into two parts; in the arrangement of Fig. 43 the outlet chamber is a metal rectangular cistern divided by a vertical partition into two compartments; the filtered water from the drainage channel enters the first compartment, from there it overflows over the partition into the second compartment, and from it is conveyed through a pipe to the clear-water reservoir. The partition dividing the cistern is made of separate sheets, and its height may be changed. The water level in the first compartment of the cistern remains constant, and the quantity of water entering the filter is also constant; therefore, when the resistance of the filtering layer to the percolation of water through it increases, the water entering the filter will at first be retained above the sand until its level rises enough to overcome the increased resistance to percolation through the filter, and then the former filtration rate will be restored. In order that the head should not exceed the greatest value,

set for the given filter, an overflow (idle) pipe is arranged at the highest permissible level.

Another method of regulating constancy of discharge, used, for example, in Moscow at the Rublevskaya station (Fig. 44), consists in the gradual automatic lowering of the water level in the outlet chamber; the water level on the filter itself is kept constant by means of an overflow pipe. The constant rate of outflow of the filtered water from the outlet chamber is achieved by the fact that the intake end of the pipe is suspended from a float, and therefore the inlet opening of the pipe is always at a definite depth below the water level in the outlet chamber; the quantity of water flowing through the opening, as is known, is fully determined by the size of the opening and its depth below the water level.

The construction of the end of the outlet pipe is shown in Fig. 45, where the regulating device of the Warsaw filters is presented. This device consists of a copper vertical movable pipe (621 mm in diameter), to the upper end of which two floats are attached (with a total area of 1 sq. m.). In the pipe, on both sides, at a height 5 cm below the water level, openings are cut, 8 cm high and 44.5 cm long. The length of the openings can be regulated by means of annular slide valves. The position of the slide valves is indicated on a scale, and the device can be set for any water discharge.

The movable pipe surrounds a vertical fixed branch of the outlet pipe; in order to reduce friction when raising and lowering the movable pipe, instead of their close contact over the whole surface, a copper grooved stuffing box is used. The movable pipe with the floats is balanced by applying weights so that the openings through which the water pours are always at an exactly defined depth below the surface of the water. In earlier-built filters, the regulating apparatuses were balanced by means of a chain passing over a block and provided with a counterweight.

In Warsaw there are no special devices for achieving constancy of the water level on the filter; therefore, when the film thickens, a lowering of the water level in the outlet chamber and a rise of it on the filter may occur simultaneously.

The admission of water to the filter after cleaning is carried out at a very low rate and in the direction from below upward (through the drainage channels), so that the air from the sand layer can gradually escape without disturbing the sand and without disrupting the uniformity of its layer. Water is supplied to the drainage channels through the outlet chamber from filter sections operating at that time.

Filter accessories. We have become acquainted with the arrangement of the devices for admitting raw water onto the filter and discharging the purified water from it. Besides these, at each filter section the following are necessary:

Fig. 45.

Fig. 46.

Fig. 47.

1) An overflow (“idle”) pipe, ensuring that the highest permissible water level on the filter will not be exceeded.

2) A drain pipe for emptying the filter when it is being cleaned; this pipe usually comes out of the filter’s outlet chamber and is provided with a gate valve (see Fig. 34).

To accelerate the emptying of the filter, a drain is sometimes also arranged at the level of the surface of the filtering layer.

3) Ventilation channels are arranged to provide free escape for the air driven out of the filtering and supporting layers and drainage channels when the filter is filled with water from below, and also to provide air with free access to the lower layers of the filtering material when the filter is emptied.

These channels are usually arranged in all the walls of the filter, at a distance of 3–5 m from axis to axis, with a cross-section of 12 × 12 cm (i.e., half a brick); they communicate with the layers of gravel lying directly beneath the layer of coarse-grained sand, as a result of which the air filling the spaces in the gravel finds for itself a free passage to the air located above the sand in the filter.

Communication between the air located above the surface of the filter and the outside air (in closed filters) is established through light wells arranged above the filter covering (at mutual distances of about 10 m, center to center). Fig. 46 shows the light and ventilation well of the Warsaw filters; the cast-iron cap is covered with a cast-iron lid, raised to illuminate the filter. The cap lid is more often made of glass for lighting.

To strengthen the ventilation of filters, channels communicating with the outside air are sometimes arranged at the upper and lower ends of the filter in the outer walls. Fig. 47 shows the supply ventilation channel of the Warsaw filters; the inflow of air is regulated by a damper; at the opposite end of the filter, brick ventilation pipes are arranged for removing air.

Open and closed filters differ in construction only in that the latter are covered and their covering is covered with earth, whereas the former are not covered.

The advantage of open filters is their lower cost; their main disadvantages are the complete lack of protection of the water from external contamination, the rapid development (in summer) of algae, leading to rapid clogging of the filter, and the warming of the water in summer and the formation of ice in winter.

The formation of ice makes cleaning the filter extremely difficult and greatly increases its cost; moreover, the sand of the filter may also freeze, especially during cleaning of the filter, which leads to disruption of the regular operation of the filter. In cases of deep freezing

water on the filter, observed near the walls of the filter, rare also because of damage to the filtering film.

A. Hazen considers it necessary to install covered filters wherever the average January temperature is below 0°.

Operation of English filters.

Having become acquainted with the arrangement of the separate parts and accessories of English filters, let us follow in detail the course of their operation and all the maneuvers necessary for managing them.

We shall consider the instructions for operating the English filters at Müggelsee for the city of Berlin.

These filters are shown in Figs. 40 and 48, and the arrangement of the inlet and outlet devices in Fig. 49.

Raw water is first admitted into the inlet chamber through the opening of pipe H (Figs. 40 and 49), regulated by a float, and from this chamber passes to the filter through a pipe ending above the surface of the sand in two mouths.

The purified water from the drains enters the collecting chamber; from it it passes into the regulating chamber (serving, as we shall see below, for regulating the water consumption and the loss of head during filtration), and only from there into the outlet chamber, from which it goes through the discharge pipe y into the reservoir of clean water.

Let us give the contents of the instructions for operating the filters at Müggelsee in Berlin, issued in 1895.

Instructions for operating the filters.

§ 1. Filling the filters. Bringing a prepared filter into operation begins with its filling, which consists in removing the air contained in the pores of the sand by introducing water through the bottom of the filter. This water is taken from the reservoir of filtered water and enters through pipe A (see Fig. 49); it enters the chambers in the direction opposite to that indicated on the drawing by the arrows, and passes into the regulating chamber through sluice valve B. Then sluice valve C is opened, but only so much that 10 hours are required to fill the filtering basin with water up to the level of the sand; float M, placed in an open vertical pipe communicating with the filter, indicates the gradual rise of the water in the sand layer.

If water were to enter the filters too rapidly, spaces filled with air could form in the sand layer, which would hinder the proper course of filtration and subsequently could cause movements (breakthroughs) of the sand.

When the water rises 0.10 m above the level of the sand, sluice valves B and C are closed, and sluice valve E is opened, through which un-

filtered water; the gate valve E should be opened slowly, so that a strong stream of water does not disturb the surface of the sand; only after the water has risen to 0.50 m above the sand is the gate valve E opened fully, in order to accelerate the inflow of water.

The water level on the filter must never rise faster than 0.12 m per hour, so that the velocity of the water’s movement is not excessive.

The inflow of water is regulated until it reaches the level of the overflow pipe. The automatic valve H regulates the filling of the filter.

§ 2. Course of filtration. The prefiltered water is allowed to be used for consumption only 24 hours after the filter has been filled, when the pores of the sand have had time to fill with silt. If this is impossible¹³, then, at the very least, the rate of flow of the filtered water should be increased very slowly. A gradual increase in the rate of flow is achieved by slowly raising the gate valve C; the water level in the regulating chamber should at first stand no more than 0.04 m above the spillway y, then it is raised little by little and, finally, reaches the normal height of 0.19 m. At this time the filter reaches its normal rate of flow, equal to 2.5 cubic meters per day per 1 sq. m of sand surface.

Having opened the inlet gate valve E, the worker must see to it that the automatic gate valve H constantly maintains the water in the filter at the level of the overflow pipe²⁶; at the same time, by lowering or raising the gate valve C, the worker causes the water level in the regulating chamber to remain 0.19 m above the spillway.

Changes in the water level on the filter, in the collecting chamber, and in the regulating chamber are indicated by means of the floats M, N, and P, on a scale fixed to the wall of the building.

It is very important to guard against any sudden changes in pressure, which, being transmitted to the layer of silt gradually deposited on the surface of the sand, immediately reduce the transparency of the prefiltered water. The same effect is produced by any sudden change in the rate of flow of the water passing through the filter. To avoid this danger and to ensure a uniform rate of flow, the worker watching the filter must continuously see to the maintenance of a constant level in the regulating chamber.

When the water has reached the normal level, the resistance to the percolation of water through the sand increases properly and gradually with the increase in the thickness of the layer of silty deposits on the filter; this makes it necessary gradually and methodically to raise the gate valve C.

During normal operation of the filter, this gate valve should not be raised in such a way that the difference between the water levels on the filter and in the collecting chamber exceeds 0.50 m; this difference, initially equal to zero, gradually increases; when it reaches 0.50 m and the normal flow through the spillway \(y\) cannot be maintained, the operation of the filter must be stopped and it must be cleaned.

After several weeks of operation of the filter, the organic substances retained on its surface may decompose and in summer begin to rot. Passing water through this rotting material is dangerous. A sign that fermentation has begun is the spongy masses floating up to the surface of the water being filtered; these masses rise from the surface of the sand and consist of a mixture of various substances with gases. When they appear, the operation of the filter must be stopped and it must be cleaned.

§ 3. Emptying and cleaning the filter. First the outlet gate valve is closed; \(A\) is opened, as well as the gate valves \(FF\) located on the overflow pipe that reaches the level of the sand in the filter. When the water drops to this level, \(B\) and \(G\) are opened; thus the water remaining in the layer of sand drains away, and the filter is emptied to the bottom, while neither the surface nor the mass of sand will be disturbed by strong jets of water. As water is removed from the sand and the supporting gravel, air fills the voids; to facilitate the access of air, vertical pipes are arranged along the walls of the filter. When filtration is carried out in full accordance with the instructions of § 2, the silty deposits do not penetrate deeply into the sand and remain almost entirely on the surface. Therefore, to clean the filter it is sufficient to remove the sand to a depth of 0.01 m.

Cleaning the surface of the filter requires experienced workers; particular care must be taken not to remove too deep a layer of sand, for the cost of cleaning depends on this. The sand removed from each part of the filter, covered by a vault and having an area of 19 sq. meters, is gathered into a pile in the middle of that part; when all the piles have been collected, they are carted in wheelbarrows to a special dumping place. The surface of the remaining sand is leveled with an implement resembling a wooden rake without teeth, taking care to furrow the compacted surface as little as possible.

The surface of the sand is made with a slope of 0.06 m between the valves \(F\) and the opposite wall, in order to facilitate the removal of water; attention must be paid to preserving this slope. When the filter has been emptied, it should, if possible, be left for some time in contact with the atmospheric air; it is left empty and, if there is no frost, the lighting wells in the vaults are left open to ensure ventilation. In winter, however, to avoid the formation of ice, the glass covers of these wells are covered with boards so as to hinder the escape of heat.

Of the eleven filters in each of the four sections, eight are sufficient, at their normal output, to feed the water-lifting pumps; thus, under normal conditions, 8 filters operate in each section, while 3 serve as reserves.

To avoid unnecessary movement of the boards on which the wheelbarrows roll, the cleaning process must be carried out consecutively, in a strict order determined in advance.

§ 4. Renewal of the sand. The sand removed during cleaning is replaced with new sand only after several cleanings, since it would be difficult to lay in place a thin layer of sand, and the operation of the filters would have to be interrupted frequently for a longer time.

The filtering action of the layer of sand gradually weakens as its thickness decreases, although the filtration process proceeds satisfactorily even when the initial thickness of the sand layer (0.60 m) is reduced by half. Therefore, replacement of the sand is carried out when the thickness of the layer has decreased to 0.20 m. For removing the sand, a small party of workers should be used; the smaller the number of workers, the shorter the stoppages of each of them at the scales when coming and going.

From time to time the side walls and the columns supporting the vaults of the basins must be cleaned of the organic deposits on them. The cleaning is done with water, from rafts or small boats. These deposits on the walls promote the development of microorganisms that may impair the quality of the water; therefore they must be removed.

§ 5. Washing of the sand. The sand removed from the filters is washed and can be used again. The washing must be so thorough that a sample of washed sand, placed in a vessel with clear water, does not cloud the water at all. For this the following precautions should be observed:

a) The duration of washing should be regulated according to the coarseness of the sand and the quantity and character of the silty deposits, by reducing the number of revolutions of the washing drum per minute while maintaining a constant flow of wash water.

b) At the same time, the water level in the washing drum should be raised.

Inorganic impurities, constituting silt, are easily removed by washing. But organic substances, such as, for example, algae, cause serious difficulties; they are hard to remove if the wet sand taken from the filters is not washed immediately but is left in heaps at a temperature favorable for the development of these plants. Thus, dirty sand should, if possible, be washed at once after its removal from the filter.

Cleaning of Filters.

When the head of water required for filtration reaches, in any compartment of the filter, the limit permitted for that filter, the admission of water into this compartment is stopped, the water is let down (sometimes entirely, but usually only to a level about 30 centimeters below the surface of the sand), and the filter is cleaned. This cleaning consists in removing the contaminated surface layer of sand, to a depth of 1–2 cm, where the greatest number of bacteria accumulate.

Cleaning is most often carried out by hand, by means of shovels, with laborers walking carefully over the sand. The dirty sand is gathered into heaps and taken away in wheelbarrows (over rolling boards), or in small cars, for washing, or to a dump; the latter is done in localities where sand is cheap.

The technique of cleaning filters has made great progress in America in recent years.

In Philadelphia, on filters built in 1900, the dirty sand is removed from the filter by portable ejectors. The ejector (Fig. 50) consists of an iron car, in the form of a funnel, placed near the laborer, into which the dirty sand is thrown with shovels; the sand first falls onto a screen, and then under a jet of water (from the supply pipeline), which carries it into the discharge pipeline leading to the place where the sand is washed. The screen can likewise be washed with water from a one-and-a-half-inch pipe connected with the supply pipeline. Along the entire filter two pipes are laid—the pipe supplying water under pressure (\(d = 2 \tfrac{1}{2}''\)) and the pipe carrying off the water with the sand (\(d = 3''\)); on these pipes, near the places where the ejector is set up, branches are arranged, to which the ejector is connected by means of rubber tubes. The ejector weighs 72 kg and can remove 3–4.5 cubic meters of sand per hour (when served by three workers).

This method has given very good results and is also used in Washington, where, since 1909, ejectors have likewise been employed to return the washed sand to its place on the filter.

A very successful method has been used since 1907 in Brooklyn (Fig. 51). When cleaning the filter, the admission of water to the filter is stopped and the water is lowered to a level several inches above the surface of the sand; then the drain pipes, located somewhat below the surface of the sand (on the right in the drawing), are opened, and through them the water that had stood above the sand flows away. Then the taps of the washing pipe are opened (on the left in the drawing); the wash water is distributed along the channel running the length of the filter, overflows in a strong stream over its edge onto the surface of the sand, and flows over it in a horizontal current. At this time the laborers, standing on the sand, sweep its surface with brooms; all the dirt is carried by the water into the drainage pipes.

Fig. 50.

Fig. 51.

Fig. 52.

Finally, in very recent times two more methods have been used for cleaning the sand in place, without carrying it far from the filter.

In the Nichols method, used since 1909 at some filters in Philadelphia, an ordinary portable ejector discharges the dirty sand into a “separator,” which likewise moves along the filter. The separator (Fig. 52) consists of a closed cylinder with a conical bottom, in which a discharge pipe (with a valve) is fixed for the sand to leave the separator after washing. Inside the cylinder there is a system of stirrers. The sand enters the separator from above and meets a jet of wash water entering the separator from below upward; the quantity of water is calculated so that the sand is not carried by the water into the overflow pipe, the opening of which, moreover, is protected by a screen. The sand comes out (at the bottom of the cylinder) completely clean. The separator moves on wheels along rails laid along the filter.

Let us point to the Blaisdell method, which has given such favorable results on the small installations where it has been used, and on the experimental installations of New York, that it is now proposed to introduce it on the filters of the Croton water-supply system in New York, which will be the largest in the world.

Blaisdell’s apparatus consists of a box turned bottom upward, which is placed under the filter water against the surface of the sand and is held in a definite position (and also moved) by means of a movable platform resting on the stone longitudinal walls of the filter. The platform moves along the filter, and the box along the platform (i.e., across the filter), as well as up and down; all movements are produced by electricity and are controlled and monitored by one person. In the box there are rotating hollow horizontal rods, with hollow vertical teeth on them, provided with openings; the axis of rotation of the rods is also hollow, and through it water is supplied under pressure. Thus the arrangements inside the box are similar to the arrangements for washing sand in an American filter of the Bell system (see Fig. 70), with which we shall become acquainted below.

The box moves over the filter, the rods rotate about their axis, and their teeth, lowered to a certain depth into the sand, loosen it while at the same time washing it with fine but powerful jets of water issuing from the openings in the teeth. A centrifugal pump is connected to the box (at the top), sucking out the polluted water in a quantity slightly greater than the clean water supplied at the same time for washing the sand; the water from the pump is carried off into the waste pipes.

The interval of time between successive cleanings of the filter depends, as we have already mentioned, on the quantity of water being filtered and especially on its quality. Thus, for example, in London in 1884, with...

it was necessary to clean the filters after passing, through 1 sq. meter of sand surface, from 40 to 127 cubic meters of water, whereas in 1892, after the enlargement of the settling basins, cleaning took place after 68–147 cubic meters had passed. In various cities of Germany, cleaning is carried out after passage through 1 sq. m from 26 cubic meters of water (Bremen) to 66 cu. m (Stuttgart). With an average filtration rate of 0.10 m per hour, the interval between cleanings in Germany is 11–28 days.

In Warsaw the cleaning of filters is performed 15–20 times a year²⁷.

Ripening of the filter. After each cleaning of a filter it is necessary, as we already know, to discharge the filtered water into the drain, owing to the unsatisfactory character of the purification, until the filter has ripened; about 12–24 hours are required for the ripening of a filter. At the same time, in order to monitor the operation of the filter, it is necessary to carry out analyses of the water until the filtering film has been restored.

When, after several cleanings, the thickness of the sand layer has significantly decreased and has reached the minimum established for the given filter (the minimum is established on the basis of the results of bacteriological analyses)²⁸, new sand must be added in order to restore the original thickness of the filtering layer.

After this, the ripening of the filter requires much more time than after cleaning: according to Riefke, the filtered water must be discharged into the drains during the first 6 days of operation of the filter.

The ripening of the filter proceeds the more slowly, the cleaner the water; therefore, in order artificially to accelerate the formation of the film, sulfurous alumina or iron salts are sometimes added to the water to be filtered, which form a sticky layer on the surface of the sand.

At the beginning or resumption of operation of a filter, it is filled with water from below, through the drains, so that the air filling the spaces between the grains of sand can escape; the water should be admitted slowly, so that the blows of the jet do not disturb the uniformity of the sand layer.

In order to avoid simultaneous replacement of the sand in several compartments of the filter, it is useful, when constructing it, to place the sand in the different compartments to different depths. In Philadelphia the Belmont filters were loaded with sand to depths of 28, 31, 34, 37, 40, and 43 inches.

Let us also note that the loss of time for cleaning filters is almost the same in winter and in summer, i.e., in winter the filtering film forms more slowly, while in summer, although the filter ripens more quickly, it also clogs more quickly, since microorganisms develop more vigorously and more frequent cleaning is required.

Fig. 53a.

Fig. 53c.

After certain intervals of time (10–15 years), all the sand of the filter and the supporting layers are removed, the entire filter compartment is subjected to ventilation and sterilization, and it is charged with new sand.

Washing the sand.

When the prices of clean quartz sand needed for charging filters are high, it is advantageous, when reloading filters, to lay again the old sand, cleaned by washing.

Usually, in the yard of the filtration station special devices are installed for washing sand with water under pressure. Washing is carried out either directly by a jet of water from a hose directed onto a heap of sand in a stone pit, over the edge of which the water overflows (Antwerp), or else by washing drums or Körting ejectors. The more primitive the washing method, the more water is consumed for it.

In Fig. 53 is shown a device used in Zurich, consisting of a washing drum and a separator. The drum, usually inclined, rotates about its axis (in Zurich by means of a water engine); dirty sand enters from one end, and water from the other end of the drum, and they remain in prolonged contact thanks to partitions arranged helically inside the cylinder and not completely blocking the cross-section of the cylinder; during rotation these partitions catch and move the sand.

On leaving the drum, the sand falls onto a coarse sieve, on which it is again sprinkled with water; a considerable part of the sand passes through this sieve; it is desirable to sort out the gravel, and for this purpose the sand that has not passed through the sieve enters a separator (a drum with sieves of different coarseness), where by sieving it is sorted by particle size.

The cost of washing 1 cubic meter of sand in Zurich is 1 fr.—1.30 fr. (i.e., 38–50 kopecks).

In Figs. 54 and 55 is shown an ejector apparatus for washing sand. Dirty sand from the filters is brought onto an inclined concrete trough, from which it is gradually lowered with shovels into the first metal funnel; into the same funnel, washing water enters from above through a pipe, diluting the sand. Ejector tubes are installed in the funnel (Fig. 55); water under pressure enters through the lower (narrow) tube and carries with it, into the upper tube, sand from the funnel. The sand, together with the water, is discharged onto an inclined metal chute ending above a second ejector funnel, of exactly the same construction as the first. All the funnels, except the first, have side spillways, through which the dirty water overflows into the discharge chute.

As the grains of sand move with the jet of the ejector, they are vigorously washed; the dissolved dirt, being lighter, floats up and pe-

pours out through the spillway, while the grains of sand settle to the bottom of the funnel, from which they are again seized by the jet of the next ejector. From the last ejector the sand emerges completely cleaned.

An ejector washing apparatus is installed, among other places, in Hamburg, where, for 18 filters with a total area of 13.70 hectares, there are 4 washing chambers, with one apparatus in each; the capacity of each chamber is 4 cubic meters of sand per hour, with a consumption of 15–20 cubic meters of washing water per 1 cubic meter of sand; the cost of cleaning is about 40 kopecks per 1 cubic meter. Three workers are required to service each apparatus.

Cost of Construction and Operation of English Filters

The cost of constructing sand filters varies greatly depending on the quantity and properties of the water to be filtered, the method of covering the filters, the prices of labor and materials, and local conditions.

The permissible filtration rate, and consequently the filter area required for the purification of a given volume of water per day, depends on the quality of the water. Therefore, the cost of constructing filters is more correctly determined per square meter of their filtering surface, and not per unit volume of filtered water.

The cost of operating filters, apart from the influence of unit prices for materials and labor, also depends chiefly on the character of the water being filtered; a substantial part of the operating expenses consists of the costs of cleaning the filters, while the intervals between successive cleanings depend on the properties of the water and its impurities.

Let us give the cost of constructing some English filters (the cost of the filtering basins, not counting the pumps, external pipelines, etc.), referred to 1 sq. m. of filtering surface.

Type of filters	City	Cost per 1 sq. m. of filtering surface
Open filters	London	12–20 rub.
Open filters	Hamburg	16 rub.
Open filters	Neuilly-sur-Marne	23 rub.
Open filters	Poughkeepsie (New York)	36 rub.
Open filters	Herdson (New York)	34 rub.
Closed filters	Berlin	32 rub.
Closed filters	Zurich	46 rub.
Closed filters	Philadelphia (Belmont)	75 rub.
Closed filters	Warsaw	38–48 rub.

(In Zurich the cost of preliminary filters is also included in the cost of the filters; in Philadelphia the cost of two

large settling reservoirs, which make it possible to increase considerably the output per square meter of filtering area).

The cost of operating filters, not counting interest and the amortization of the capital expended on the construction of the filter, referred to 1 cubic meter of filtered water²⁹, in Germany ranges between 0.08–0.16 kopecks, rising in rare cases (Altona) to 0.25 kopecks; in England: in Liverpool about 0.06 kopecks; in London the average cost for the period 1880–1895, for eight companies supplying water to different parts of the city, ranges from 0.04 to 0.08 k.; the difference in cost in individual years is very considerable (sometimes in one year the cost is twice that of another year) and increases especially with the appearance of ice on open filters.

In general, the cost of operating open filters is higher than that of closed ones.

The cost of operating the filters of the “Compagnie Générale des Eaux” in France in 1897 was expressed by the following figures (as before, per 1 cubic meter of filtered water):

Operation of filters	0.43 cm	0.25 cm	0.15 cm
Servicing of water-lifting machines	0.20 cm	0.15 cm	0.02 cm
Interest and amortization of capital	0.47 cm	0.40 cm	0.43 cm
Total	1.1 cm = 0.44 k.	0.8 cm = 0.32 k.	0.6 cm = 0.24 k.

Operation in Zurich, including interest (4%) and amortization (2%), costs: with open filters 0.37 kopecks and with closed ones—0.34 kopecks.

Finally, in America the average cost of operating filters (without machines, interest, and amortization) varies from 0.09 to 0.16 kopecks (the latter figure refers to open filters in the event of frosts).

Advantages and disadvantages of English filters.

English sand filters, at a low filtration rate, with proper construction and maintenance, give a high degree of water purification, sometimes reaching retention of 98–99% of the bacteria contained in the water; they remove turbidity in the amount of 50–75 parts by weight per million parts of water (i.e., 50–75 mgr. per liter) and reduce the coloration of water caused by organic (vegetable) admixtures by 20–30%. So long as the composition of the water being filtered is not subject to sharp

changes (usually occurring in river water in spring), the results of the action of sand filters are not subject to noticeable fluctuations, and therefore the quality of the filtered water may be considered constant.

Among the shortcomings connected with the action of sand filters¹⁴, we shall indicate the following as essential:

First, filtration proceeds slowly, and as a result a considerable area is required for the construction of filters.

Second, the cleaning of a contaminated filter requires much time and labor, and the contact of the workers’ feet with the material being filtered—something that has not been eliminated even in the improved American methods of cleaning—cannot be considered desirable from the sanitary point of view.

Third, at each cleaning of the filter there is a considerable loss of time until the film is restored, as a result of which the filter proves to be out of operation for 10–20% of the total time.

Besides the loss of time, cleaning is associated with a disturbance of the normal operation of the filter and therefore requires intensified supervision of it and more frequent bacteriological analyses of the filtrate. Hence the understandable desire to increase the intervals of time between cleanings by means of preliminary rough clarification of the water.

Fourth, English filters do not possess the ability to deprive water of color in general, and of opalescence in particular, producing only a weakening of color by 20–30%; meanwhile, in many cases, especially in water supply from rivers fed by swamps, this is an urgent necessity.

In view of the above, English filters, without any additional treatment of the water, are applicable especially for the purification of waters containing very few vegetable impurities and color, and comparatively little turbidity—that is, for water almost transparent and colorless in appearance—which these filters bring to a high degree of purity.

Let us note in conclusion that filtration through English sand filters, as through any other filters (i.e., porous substances), while retaining bacteria mechanically, i.e., as the finest suspended particles, cannot ensure the absence of pathogenic bacteria in the filtered water, since among those 5–1% of bacteria that remain in the filtered water, pathogenic ones may also be found. Therefore, in cases where there is reason to fear the presence of pathogenic microorganisms in the water (for example, during epidemics), the water, even if well filtered, must still be subjected to disinfection, i.e., treatment by such methods as act specifically upon bacteria as living organisms and kill them.

Chapter IV

Modifications of the English Method of Filtration

In addition to the use of slow sand filters in their original form, certain modifications of them are often employed, which arose as the result of observations of the process of their operation.

The modified methods of filtration through sand may be divided into three groups:

a) the filtration process is divided into several stages;

b) in addition to the mechanical action of the filter, the biological process taking part in it is intensified by supplying atmospheric oxygen in greater quantity than in English filters;

c) the formation of the film is artificially accelerated by the addition of chemical reagents, thereby reducing the loss of time for cleaning the filter; this process is used in the so-called “American filters.”

Filtration with Prefiltration

We have seen that in English filters the content of bacteria and suspended matter is reduced chiefly by the surface of the filter; hence the idea arose of passing the water twice through the surface of the filtering material, but at a higher rate, i.e., in place of one filter, to arrange two successive filters of smaller area.

With such an arrangement the second filter must retain those particles that have passed through the first; consequently, it must consist of finer material than the first.

The first filter (called the “preliminary” filter—prefilter, for-filter, or “coarse”—“roughing”) is made of sand of greater coarseness than ordinary English filters, while the second (“final”) filter is arranged according to the usual type. The prefilter frees the water from large suspended particles that would quickly clog the fine sand of the filter; thus, in addition to the double action of the surface layers on the water being filtered, we obtain the possibility of cleaning the fine-grained filter less frequently. With this last-

from the standpoint of its operation, prefiltration is a development of preliminary settling, which is also not infrequently applied before water purification on English filters or prefilters.

The installation of prefilters makes it possible to increase the rate of filtration on the final (English) filter from 2.4 m to 5–7 m per day.

From the economic point of view, in many cases (depending on the character of the water) it proves cheaper to purify water to an equal degree of purity on filters with prefilters than on English filters alone.

Double filtration through successive filters of equal area has been used, for example, in Bremen¹.

However, the installation of prefilters of the same area as the filters may be recommended only as a temporary measure at an existing filtration station, when the composition of the raw water suddenly deteriorates so much that its purification by English filters proves unsatisfactory (for example, in spring, when the water carries much turbidity); then the existing filters can be divided into two groups, pumping the filtrate of the first group onto the filters of the second group for final purification and passing the water through each filter at double the rate: such double filtration, with very turbid water, generally gives better results.

However, in permanent installations the prefilters should be made of smaller area and with a coarser charge than the final filters.

There have been quite a few cases when the addition of prefilters to existing filters greatly increased their productivity and the quality of purification.

The results of the operation of prefilters are evident from the example of the city of Zurich, whose water supply we shall consider in somewhat greater detail.

In Zurich the water is taken from the lake, at a distance of 300 m from the shore, by means of a pipe 0.90 m in diameter, which can deliver 48,000 cubic meters per day; the filtration station is located in the city, 2 kilometers from the water-intake point.

The filters (Figs. 56 and 57) are covered, 10 in number, each with 675 sq. m of useful area. Each filter compartment has a separate prefilter, with an area of 70 sq. m (i.e. about 0.10 of the filter area), located above the middle part of the filter. The filters are built of concrete; the prefilters, built later, are made of reinforced concrete (the Hennebique system).

The prefilters are filled with gravel of 2 mm size, with a layer thickness of 0.90 m; above the gravel is poured a layer of fine sand 0.03 m thick. The depth of water on the prefilters is 0.45–0.50 m above the sand.

Fig. 56 and 57.

Fig. 59.

Final filters contain a layer of sand 0.90 m thick, supported by gravel which, together with the drainage channels, occupies 0.50 m in height. The water level on the filter is automatically maintained 1 m above the surface of the sand by means of a regulator with a float.

The inflow of water to the prefilters is regulated by hand, by means of a disk raised and lowered in a conical inlet pipe.

The filtration rate on the prefilters is 50–70 meters per day, and on the filters from 4 to 7 meters (on average 5 m.; a rate of 7 m. occurs when water consumption reaches its maximum).

The prefilters are cleaned with water let in in the reverse direction to that used in filtration, i.e. from below upward, through special tubes lying on the bottom of the prefilter; at the same time, for better mixing of the wash water with the sand, compressed air is forced into the lower layer of sand through openings in special tubes likewise laid along the bottom of the prefilter. The dirt is carried by the wash water upward, whence it is discharged into the drain pipes; when the sand settles and beds down, the prefilter is again ready for operation.

The prefilters retain about 90% of suspended matter and 50% of bacteria; they require cleaning approximately every two days, and this takes 20–30 minutes. The material of the prefilter (gravel) is renewed once or twice a year.

The introduction of prefilters increased the intervals between successive cleanings by 3–4 times.

The results of the operation of filters with prefilters are excellent. Analyses of raw water are made every two weeks, and of filtered water twice a week and more often (in the event of a deviation from the normal composition); before the cleaning of the English filters and after it, daily analyses are made.

Average composition of the water before filtration:

Component	Amount
CaO	56.5 mgr. per liter
MgO	9.8 mgr. per liter
Alkalis	4.5 mgr. per liter
Iron and alumina	0.9 mgr. per liter
Silica	1.9 mgr. per liter
Chlorine	3.2 mgr. per liter
Sulfuric acid	11.6 mgr. per liter
Nitric acid	1.8 mgr. per liter
Carbon dioxide in various compounds	48 mgr. per liter

The temperature varies from 3.8° (in January) to 12.9° (in August).

Plankton, consisting chiefly of algae, weighs on average, after drying, 1.065 g per 1 cubic meter of water.

The number of bacteria fluctuated, for example, in 1900 between 62 and 18,557 in 1 cubic cm, with an average quantity of 2,000.

Filtration reduces the number of bacteria to 7—132 (on average to 37) in 1 cubic cm.

The oxidizability of the water (characterizing the content of organic substances), equal on average to 4.36 mg (of oxygen per liter of water), decreases after filtration to 2.78 mg.

Protein ammonia decreases from 0.080 mg to 0.018 mg; finally, free ammonia, which is contained in the raw water in an amount of 0.003 mg, is no longer encountered.

The diagram in Fig. 58 shows the composition of the water for 1902 before treatment, after prefilters, and after filters. From it it is evident that the role of the prefilters is especially important for retaining plankton, which without prefiltration would quickly clog the filters. Prefiltration, by clarifying the water, allows the final filters to carry out their chief task—retaining bacteria—especially intensively.

In the example of the city of Zurich, the total required useful area of filters and prefilters per 1000 cubic meters of water per day is equal to

\[ \frac{1000}{50}+\frac{1000}{5}=20+200=220\ \text{sq. m,} \]

whereas with ordinary English filters, at the normal rate of filtration, there would be required

\[ \frac{1000}{2.4}=416.5\ \text{sq. meters.} \]

A large installation of filters with prefilters exists in Moscow, at the Rublevskaya water-lifting station of the Moskvoretsky water-supply system.

Water from the Moskva River, before filtration, is first subjected to sedimentation; the length of the water’s path in the settling basin is 25 sazhens, the velocity of motion 2 mm per second. In spring, when the water carries much suspended matter², before the water enters the settling basin a solution of aluminum sulfate is sprayed into the water conduit, for better precipitation of the suspended matter.

The preliminary filters are intended to lengthen the operating period of the English filters, which become clogged very quickly with turbidity, and especially with the flocs formed when aluminum sulfate is added.

The preliminary filters, numbering 16 compartments, were built in 1906 of reinforced concrete and are designed for 4 million buckets per day,

at the maximum filtration rate of 1.5 m per hour. The dimensions of each compartment are: length 8.5 sazhen, width 2.5 sazhen, depth of water above the bed 2.13 m (= 1 sazhen).

The bed of the prefilters consists of five layers of different grain size, namely (from bottom to top):

1) gravel of grain size 16–25 mm, with a layer thickness of 8 cm.
2) ” ” 10–15 mm, ” ” ” 7 cm.
3) ” ” 6–10 mm, ” ” ” 10 cm.
4) ” ” 3–6 mm, ” ” ” 10 cm.
5) coarse sand 1–3 mm, ” ” ” 40 cm.

The prefilters are cleaned by passing water from the settling basin upward from below and by blowing air compressed to a pressure of 1.5 m of water column. The quantity of air supplied by the compressor per minute is 55 cubic meters. The blowing of each compartment continues for 30 minutes. The air, penetrating the bed, stirs up on the surface of the sand the deposits, which are then carried away by the current of wash water into the drain.

From the prefilters the water passes to English filters, consisting of 10 compartments, with a total area of 28,820 sq. m, certain details of which we have already discussed above (Figs. 38 and 44); their design filtration rate is 10 cm per hour, although at present the usual rate is lower, since the demand for water has not yet reached the full design value.

Let us also mention the prefilters in Nice, where they are given a form different from the usual one: the prefilters (Fig. 59) consist of vertical elements with double walls (of concrete) with circular openings in them; sand is filled into the spaces between the walls. Water enters (from the supply pipe $A$) along channel $B$ into the spaces $C$, whence it seeps (in a horizontal direction) through the filtering slabs into the spaces $D$, from which it exits along channel $E$ into the pipeline to the filter.

The principal convenience of vertical filtering elements is the attainment of a large filtering surface with small plan dimensions of the basins; but their drawback is the uneven rate of water seepage through the slabs at different depths. For prefiltration, however, this drawback is of no substantial importance. The operation of the prefilters in Nice is entirely satisfactory.

Puech-Chabal Filters (Puech Chabal)³.

A further development of the principle of relieving the final (English) filter by means of preliminary settling or coarse—

³ Filters after Puech’s idea, applied to a design developed by the engineer Chabal.

Fig. 60.

Fig. 62.

Fig. 61.

filtration consists in passing the water being purified through several successive filters, with media of different grain sizes, beginning with crushed stone and ending with fine sand.

Fig. 60 presents the scheme of a filter of the Puech-Chabal system, widespread in France and also used in London, Antwerp, Magdeburg, Tiflis, and other cities.

Raw water first enters a battery for the removal of large particles, or coarse filters (“degrossisseurs”), usually consisting of four compartments filled with gravel of a grain size of 15–30 mm in the first compartment, 10–15 mm in the second, 5–10 mm in the third, and 4–6 mm in the fourth. The thicknesses of the gravel layers are respectively 30, 35, 40, and 45 cm, but this thickness is sometimes varied according to the properties of the water being filtered. Water, understandably, passes more easily through coarser material, and therefore the rate in the first compartment is greater, while the area is smaller than that of the second, and so on.

The rate of percolation of the water decreases rapidly from one compartment to the next, and in the last is 5 times less than in the first.

The transfer of water from one compartment to another is effected by means of spillways, in the form of double and triple cascades; thanks to this the water mixes with air (is aerated), which, of course, contributes to its purification.

Action of the degrossisseur. It has been established that suspended impurities are retained throughout the entire thickness of the material in each compartment of the coarse filter. In observations on an experimental filter with an area of 10 sq. m, Dr. Kemna established that the action of Puech’s coarse gravel filters gives good results as a preliminary purification of raw water. He found that a large part of the suspended substances is retained, the number of bacteria is greatly reduced, a certain decrease in organic substances and albuminoid ammonia is observed, and a considerable diminution of free ammonia.

Kemna also established that the reduction in the number of microorganisms and suspended particles does not depend on the rate, whereas the chemical changes (a decrease in the amount of ammonia and organic substances) are inversely proportional to the filtration rate. These propositions are very important.

Prefilters and the final filter. From the last compartment of the degrossisseur, the water passes (also in the form of a cascade) to the next stage of purification, to a prefilter filled with coarse sand (grain size 2–4 mm), and from there, again by cascade, to the final sand filter (English). The thickness of the sand layer in the latter is 70–90 cm, with a grain size of 2 mm; beneath the sand are the usual supporting layers.

The rate of percolation of the water in the first compartment of the degrossisseur may reach 320 meters per day; the rate on the final

Water analyses: raw, from the prefilters and filters.

Content of organic matter, mg/L

Free ammonia, mg/L

Albuminoid ammonia, mg/L

Plankton, mg/L

Number of bacteria in 1 cu. cm.

January February March April May June July August September October November December

Fig. 58 (to p. 102).

filter—somewhat more than 3 m. Thus, the total required area of all compartments is greater than in single-stage English filters.

To regulate the rate of outflow from the final filter, a regulator is placed at it, the arrangement of which is shown in Fig. 61. The following designations are used in this drawing:
$A$—siphon; $H$—the height at which the siphon operates. It is regulated by screw $C$; $C$—a screw passing through nut $E$ and serving to support the siphon; $S$—a valve regulated by a float; $F$—a float attached to rods $Tgg$, which support the siphon; $P$—counterweight.

The intervals between cleanings of the final filter are 6—11 months; the prefilter must be cleaned every 5—20 days, and the roughing filters—every 8—28 days. Cleaning is carried out by a reverse flow of water (from bottom to top).

A large installation according to the Puech system was built in Magdeburg in 1909 for the purification of water from the River Elbe, known to be polluted with sewage and other impurities; the success achieved was complete.

In Magdeburg the daily water consumption is approximately 45,000 cubic m. The amount of dry residue in the water fluctuates from 400 to 800 mgr. per liter.

Before the construction of the Puech filters, the water, after leaving the settling basins (through which it flowed for 12 hours), entered “fore-filters” (preliminary filters), with a total area of 1,380 sq. m, consisting of a layer of gravel 60 cm thick, with grain size of 1 cm; after leaving the fore-filters, the water entered English filters with a total area of 18,330 sq. m, with a layer thickness of 1 m. The fore-filters quickly became silted throughout the entire thickness of the charge; their cleaning was very difficult, and their charge often had to be renewed. The English filters had to be cleaned every 3—9 days, when the loss of head increased from 20 to 100 cm. The results of purification were very uneven, and the number of bacteria in the purified water fluctuated from 26 to 5,800 in 1 cubic cm.⁴)

The Puech–Chabal installation, put into operation in the summer of 1909, consists of four successive roughing filters (dégrossisseurs); water is supplied to the first of them and passes from one to another in the form of triple (three-stage) cascades. The total area of the first dégrossisseur (divided into 8 compartments) is 160 sq. m, of the second—280, of the third—512, and of the fourth—1,176 sq. m. The gravel filling the dégrossisseurs is laid on perforated metal sheets, through which (as through drains) the water seeps; cleaning is carried out by a reverse jet of water and by blowing with compressed air.

⁴) Walter Clemence. “Engineering” 1910.

After the degrossisseurs, the water enters a coarse-grained sandy (preliminary) filter, with a total area of 4,000 sq. m (divided into 25 compartments), from which it comes out completely clarified. Final purification is carried out on the former English filters, with an area of 18,330 sq. m. The old settling basins and forfilters have been discarded. In passing from the degrossisseur to the preliminary and final filters, the water is again aerated (falling in cascades).

Chemical analysis of the aerated water before entering any compartment of the degrossisseur and upon leaving it showed that all the oxygen stored up by the water during aeration is consumed in the filtering layer for the oxidation of organic impurities, and upon leaving the degrossisseur the water is very poor in oxygen.

The system of degrossisseurs and the preliminary filter together reduce the bacterial content of the raw water by 90–95%, making the number of bacteria in the filtrate practically constant (20–40 in 1 cu. cm.).

In this system, the English filter works exclusively for the bacterial purification of water that reaches it completely clarified. Therefore it is not surprising that the loss of head in the final filter increases over the course of a month by only 1–4 cm.

In Russia an installation based on the Puech-Chabal system has been built in Tiflis, for clarifying the very turbid water of the Kura River. The amount of dry residue in the water is 150–230 mg per liter, and during floods reaches as much as 1500 mg per liter.

The installation is designed for 500,000 vedros (= 6000 cu. m) per day and consists of three successive forfilters, one preliminary filter, and an English filter.

The concrete rectangular forfilters each have a depth of 2.5 m; the area of the first forfilter is 50 sq. m; it is charged with gravel of grain size 10–15 mm, with a layer thickness of 300 mm; the second forfilter, with an area of 100 sq. m, is charged with finer gravel of 5–10 mm, layer thickness 350 mm; the third—with an area of 150 sq. m, with grains from 4 to 6 mm, and a layer thickness of 400 mm. After the forfilters, the water enters the preliminary filter, with a total area of 300 sq. m (divided into 4 compartments), charged with sand (1–2 mm) in a layer of 600 mm; the filtration rate is up to 85 cm per hour. The English filters are charged with a layer of sand (0.4–0.5 mm), 700 mm thick, and with two supporting layers of gravel of 200 mm each; the filtration rate is up to 30 cm per hour³. Washing of the forfilters and preliminary filters is carried out by passing water from below upward

\[ \frac{1000}{0.30 \times 24}+\frac{1000}{0.85 \times 24}\times 2=139+98=237\ \text{sq. m, whereas} \]

with successive relaying of the sand. Washing of the preliminary filters is carried out once a week; of the preliminary filters—once every two weeks; and of the English filters—once every 1–1½ months.

The costs of maintaining the filters for each 1000 buckets of filtered water amount to as much as 2 kopecks.

The roughing filters and preliminary filters retain about 50% of the bacteria, while after the English filters about 20% of the bacteria still remain in the water, so that the bacterial content in the purified water ranges from 40 to 540 per 1 cubic cm, which cannot be regarded as satisfactory. The design filtration rates are so high that during spring floods the water cannot be clarified by the filters; fine silty turbidity easily passes not only through the gravel filters but also through the sand roughing filters, and the operating conditions of the English filter are extremely unfavorable, and the filters rapidly become clogged. To avoid this, it is necessary to resort to coagulation of the water (5–8 poods of aluminum sulfate per 400,000 buckets of water), with its settling for 5 hours in settling basins; when settling tanks are available, however, the gravel roughing filters are of little benefit.

Cases of the use of double filtration and Puech filters. Filtration on English filters with prefiltration is especially suitable for the purification of waters containing many suspended (but not colloidal) impurities and plankton, occupying in quality a middle position between clear waters containing little turbidity (not more than 50–75 mg per liter), which can be subjected at once to high-quality purification on slow sand filters, and waters polluted with dissolved or colloidal impurities, the purification of which is possible only with the addition of chemical reagents (coagulants).

Intermittent filtration and unsaturated sand filters.

There is no doubt, as we have already said, that in the process of slow filtration on English filters, in addition to mechanical factors, biological factors also participate; but for their more vigorous manifestation, on ordinary filters covered with a layer of the water being filtered, the inflow of air necessary for the vital activity of most microorganisms is too small.

with ordinary English filters would be equal to

\[ \frac{1000}{0.10 \times 24} = 417\ \text{sq. m.} \]

However, the unsatisfactory results of purification point to the inadequacy of the areas adopted.

The fact has also been established that air, penetrating the filters, promotes the oxidation of organic substances in the water and its purification.

Puech, by passing the water being filtered from one compartment of his filters to the next in cascades, enriches it with oxygen, which, together with the water, is then introduced into the thickness of the filtering material; this may partly explain the greater degree of purification from organic impurities than in other filtration systems.

Finally, in the natural filtration of water through soil, air undoubtedly penetrates into the pores of the earth. The English filtration process differs from natural filtration, among other things, in that the surface of the filtering sand is covered with a layer of water (submerged).

In view of this, the idea arose, in imitation of natural filtration, of giving air access to the pores of the filtering layer, making the filters operate only periodically, or else making them unsubmerged.

Intermittent-action filters, installed experimentally in Lawrence in 1893, showed that the oxidation of organic substances and nitrification (the conversion of ammonia into nitrous and nitric acids) proceed better than in continuously operating filters, but the reduction in the number of bacteria is accomplished considerably worse. The latter is explained by the fact that, in order to achieve the same capacity per unit area of filter, intermittent filtration requires increasing the rate of passage of the water; moreover, alternating filtration and aeration damage the filtering film, and the purification of the water is accomplished chiefly by the sand, almost without the participation of the film. Be that as it may, with sand of good quality and a thick layer of it, the number of bacteria in Lawrence was reduced from 2,000–25,000 in 1 cubic cm of raw water to 100–400 in the filtered water.

The use of intermittent filtration is suitable for waters heavily polluted with organic impurities, which require contact with atmospheric oxygen. When intermittent filters are used for such waters as preliminary treatment, there is no need to construct reservoirs for the filters; one may confine oneself to laying drainage pipes beneath a layer of sand of the required quality, placed in place of the excavated soil (if, of course, climatic conditions permit). However, waters with a high content of organic impurities are best, whenever possible, not used for drinking at all.

Unsubmerged sand filters ^6) (Fig. 62) were proposed by the well-known French bacteriologist Miquel; their difference from English filters is that the water falls onto the surface of the filtering—

^6) L. Baudet. Filtres à sable non submergé. 1908.

a step of sand in the form of rain from pipes with openings, laid at a certain height above the sand.

The filter must have a layer of sand not less than 1.30 m thick, with a sand grain size of 0.5 mm; the total thickness of the drainage and supporting gravel layers was arranged at 0.22 m.

At the beginning of operation the filter retains very few bacteria, but after approximately 16 weeks the bacterial content in the filtrate is established at no more than 100 per 1 cu. cm.

No filtering film at all is formed on the surface of the sand; filtration of the water is carried out exclusively by the layer of sand, and therefore the operation of the filter is very uniform; there is no danger of disturbing or tearing the film. Cleaning of the filter was required after more than a year had elapsed.

According to the investigations of Miquel and Mouchet, in the sand down to a depth of 30 cm there are many bacteria, which disappear at greater depth.

The action of the filter on B. coli should be noted; when B. coli was present in the raw water even in considerable quantity, it was never possible to find it in the filtrate, even in tests of large volume. In the sand, B. coli penetrated only to a depth of 50 cm. When typhoid bacilli and cholera vibrios were added to the raw water, they likewise were completely unable to pass through the matured filter.

The chemical qualities of the water are considerably improved by filters of this system.

The results of laboratory experiments were fully confirmed at the large installation in Châteaudun, operating under constant supervision for several years.

The construction of non-flooded filters should not prove significantly cheaper than ordinary English filters, but their maintenance requires less supervision.

The rate of filtration cannot exceed a certain value; however, judging from the available experimental data, after the filter has matured one may safely allow a rate 2–3 times greater than on English filters (i.e., 4.8–7.2 meters per day).

Several types of non-flooded filters have already been designed, and their use is highly recommended in France¹⁷. In Germany, too, it is considered desirable to carry out further systematic experiments with them. However, these filters have not yet entered into practical use.

Chapter V

American (Mechanical) Filters

The lack and high cost in cities of the large areas of land required for the construction of English filters, and the need for satisfactory purification of water when its quality changes sharply—as happens in spring during floods—contributed to the considerable success of American (“mechanical”) filters, which are becoming more widespread every year.

In America many rivers are subject to very severe freshets and, for several months, carry an enormous quantity of silt, whose extremely fine particles often cannot be retained by English filters; meanwhile, certain branches of industry that are very widespread in America (for example, the manufacture of writing paper) require water absolutely free of suspended matter. To meet their requirements, the idea arose of adding to the water coagulants, or chemical reagents, which cause the rapid precipitation of suspended substances. With subsequent filtration of the water it becomes possible to pass it through a sand filter at a rate considerably (as much as 50 times) exceeding the rate in English filters, since the fine suspended particles, gathered by the coagulant into flocs, are readily retained by the filter. At the same time, on the surface of the filtering material a film is rapidly formed, playing a role similar to that of the filtering film of English filters.

Such is the basic principle of American filters.

The first filter of the American system was invented by Hyatt around 1883 and came into use in 1887.

Coagulation.

“Coagulants” are the name given to chemical reagents which, when added to water, combine with substances present in the water and form an insoluble, flocculent or gelatinous precipitate, which, as it settles, carries down particles suspended in the water, including fine turbidity, coloring substances, and partly bacteria.

Some portion of the coagulant and of the substances carried by it out of the water remains in the settling tank, while the remaining, finer and lighter particles pass with the water onto the filter, promoting the rapid formation of a filtering film on its surface.

For coagulation, all kinds of chemical reagents may be used that are capable of producing the above-mentioned precipitate with the water being purified, provided only that they meet the requirement of not being harmful to health and of not leading to the formation in the water (after the reaction) of harmful compounds.

Most often used for coagulation are sulfuric-acid alumina (aluminum sulfate, \(\mathrm{Al_2(SO_4)_3}\)), ferric sulfate \(\left(\mathrm{Fe_2(SO_4)_3}\right)\), alum \(\left(\mathrm{KAl(SO_4)_2}\right)\), and certain other reagents. The choice of the coagulant most suitable for the water being purified, and the determination of the required quantity of it, are made by means of laboratory experiments. The amount of coagulant added is generally very small; for example, sulfuric-acid alumina usually has to be added at 25–50 g per 1 cubic meter of water.

The reaction that takes place in the water when a coagulant is added consists in its decomposition owing to the presence in the water of carbonates of lime and magnesia; the sulfuric acid contained in the coagulant combines with the lime or magnesia, carbonic acid is released, and the oxide of aluminum (or of iron, if ferric sulfate is used), combining with water, forms a coarse, insoluble, gelatinous hydrate \(\left(\mathrm{Al_2O_3 \cdot 3H_2O}\right)\), which constitutes the coagulating agent and is precipitated in the form of flakes, which draw into themselves and mechanically carry down with them the suspended impurities.

The decomposition reaction of a coagulant, for example sulfuric-acid alumina, proceeds according to the equation:

\[ 3\mathrm{CaCO_3} + \mathrm{Al_2(SO_4)_3} = 3\mathrm{CaSO_4} + \mathrm{Al_2O_3} + 3\mathrm{CO_2}. \]

If the water is too soft, and the content in it of carbonic-acid lime (or magnesia) is insufficient to decompose such a quantity of coagulant that the flocculent precipitate formed will carry down all the turbidity, then it is necessary to add a solution of lime to the water (in the form of lime water).

Each German degree of temporary hardness of the water is capable of decomposing 0.5 g of sulfuric-alumina salt¹ per 1 bucket of water, or 40 g per 1 cubic meter of water.

When the amount of coagulant is greater than can combine with the lime and magnesia, it remains undissolved in the water, giving it an unpleasant taste; therefore its quantity must be strictly calculated.

After coagulation, the temporary hardness of the water decreases, while the permanent hardness (caused by sulfates) increases by the same number of degrees—

The addition of 1 gram of commercial ferrous-alum salt to 1 bucket of water increases the permanent hardness by 2 German degrees.

For the precipitation of fine silty impurities a larger dose of coagulant is required than for coarse ones.

The role of the coagulant is not exhausted by the capture and precipitation from the water of turbidity (including bacteria) that cannot be removed by simple settling and filtration; it also frees the water from organic substances dissolved in it. In the case of a considerable content in the water of organic substances of a humic character (from admixtures of swamp and meadow waters), which give it a yellowish tint, coagulation not only sharply lowers oxidizability, but also noticeably weakens the coloration of the water.

Mixing the water with the coagulant solution is carried out in a settling tank, the capacity of which may vary greatly, depending on the quality of the water; in Petersburg, for example, at the filtration station the settling tanks are designed for a two-hour residence of the water with the coagulant in them. A schematic drawing of the settling tank in Petersburg is given in Fig. 124: water from the supply pipe passes along a trough, where the coagulant is added to it, into a vertical pipe standing in the center of a cylindrical settling reservoir; from this pipe the water enters the reservoir near its bottom and leaves the reservoir by overflowing over its upper edge into a circular channel, whence it flows through pipes to the filter.

A coagulant solution of a definite strength is always kept in a special reservoir, from which it is supplied to the water in a strictly measured quantity; the feed is effected either through a siphon from a vessel with a constant level, as in Fig. 161², or through a calibrated tap, or, finally, in the case of constant and considerable fluctuations in the quantity of water entering the settling tank, the solution is supplied by means of scoops (which take up the solution and pour it into the water), set in motion by a hydraulic wheel, which is rotated by the water entering the settling tank.

The solution is sometimes fed through a pipe directly into the settling tank (at a height of 1.5–2 feet from the bottom), but it is better to introduce it into the channel or pipe that supplies the water to the settling tank, since during the movement of the water the solution mixes better with it; for the same purpose, it is desirable that the solution

General description of the arrangement of American filters.

The essence of the arrangement of all systems of American filters is the same, and they differ only in details.

Fig. 64.

Fig. 63.

Fig. 65.

Fig. 66.

Fig. 67.

Fig. 69.

American filters (see Figs. 63–65) consist of a reservoir (wooden, iron, stone, or concrete), with a double bottom; the upper bottom is provided with numerous small openings. Above the bottom the filtering material is poured, usually quartz sand (sometimes crushed quartz); the thickness of the sand layer is 0.75–0.90 m, sometimes greater (reaching 1.50 m), with a grain size of 0.30–0.50 mm.

The filter reservoir may be open in so-called gravity filters, in which the filtration of water takes place under the influence of the difference in water levels above the filter and at its outlet; in filters of another type, operating with water forced onto the filter under pressure and called pressure filters, the reservoir is made hermetically closed.

Pressure filters are used chiefly in those cases where one wishes to avoid double pumping of the water (before the filter and after it); however, the results of water purification in them are worse than in gravity filters.

Finally, the axis of the filter reservoir, if it is made cylindrical, is arranged vertically or horizontally.

Almost all designers adapt their system to all these conditions, i.e., they build, as desired, gravity or pressure filters, vertical or horizontal ones; filters are made in various sizes, from small domestic filters to large reservoirs several meters in diameter and height. For the water supply of cities, several apparatuses are connected in a battery, with settling reservoirs arranged separately.

In American filters, certain auxiliary parts are very important, especially the mechanical device for stirring the sand during washings, which are carried out very frequently (approximately once a day) by passing water in the direction opposite to that used during filtration, i.e., from below upward; the sand is stirred with the water by the rotation of mechanical rakes, and the dirt is carried upward by the water and overflows into the waste pipe.

The mechanical rakes (see Fig. 63) consist of rods strengthened at right angles to the vertical axis and rotating together with it; on each rod is mounted a row of teeth which enter the sand and stir it as the axis rotates.

Because of the mechanical devices for washing the sand, American filters have also received the name “mechanical.”

Of great importance is the automatic regulator of the flow rate of the filtered water (“controller”), which ensures a constant filtration rate independently of the depth of water on the filter, and is used in gravity filters.

In some filter systems the regulator (Weston system, Fig. 67) plays yet another role: it is placed considerably

below the bottom of the filter (at least 4 feet), in order to create suction from the filter, owing to which the sand, tending to settle, is compacted and filtration is improved. The level of the newly laid sand, under the influence of compaction, falls by 1 inch or more.

Weston’s regulator (Fig. 67) consists of a cast-iron box with internal partitions forming a siphon, and with a spherical copper float, which controls the valve for admitting water from the filter into the regulator, maintaining a constant water level in the regulator. With a constant level, the outflow of water from the regulator box depends only on the size of the outlet opening, which can be changed by inserting various diaphragms. Owing to the construction of the regulator in the form of a siphon, a vacuum is formed which draws water out of the filter.

A drawback of pressure filters is the absence of instruments for regulating the filtration rate, which, when water is forced directly by pumps into the filters, changes as the filtering material becomes clogged, and thus the quality of the filtrate also changes.

The filtration rate in American filters is usually 4–5 m per hour (sometimes even more, up to 8 m), i.e. it exceeds the rate in English filters by 40–50 times and corresponds to the output of 1 sq. meter of filtering surface at 96–120 cubic meters per day. The loss of head during filtration reaches 3 m (instead of 0.6–1.0 m in English filters).

The cleaning of American filters, naturally, is carried out the more often the more turbidity there is in the water being filtered. As we already know, all the sand is washed in the filter itself; about 5% of the quantity of water being filtered is spent on washing. The usual rate of passage of the wash water is about 1 foot per minute, i.e. approximately 4 times greater than the filtration rate. Cleaning, repeated about once per day, takes 10–20 minutes, is performed much more easily, and costs much less than in English filters. When the operation of the filter is resumed, at first the purified water obtained is unsatisfactory until the filter (film) is “restored,” but the restoration process requires little time (about 15 min.), and the quantity of water that has to be discharged into the drain does not exceed 3% of the total amount.

It is desirable to wash the filter with filtered water; if unfiltered water is used, then before resuming the operation of the filter a certain quantity of clean water is passed through it (in the normal direction).

Finally, the filters are sterilized from time to time (1–2 times a year, or more often if the water contains fatty or oily impurities).

Sterilization is carried out with hot steam or without it; after pouring sodium bicarbonate over the surface of the filtering material in an amount of about 2.5 kg per 1 sq. m, water is admitted in such quantity that a layer several centimeters deep forms above the surface of the material; all drainage pipes must be closed. The solution is left for fifteen hours, after which the filter is washed as in ordinary cleaning.

Sterilization by means of steam should be preferred; after adding soda and water, as indicated above, steam is admitted under pressure (through the drainage pipes, if there are no special steam pipes); the steam brings the soda solution to boiling. After 1 hour the admission of steam is stopped and the filter is washed. After sterilization, the filtrate is discharged into the sewer for 45 minutes.

Very little labor is required for the maintenance of American filters; for example, in Cincinnati 2–3 workers service the entire filtration station, which purifies about 400,000 cubic m of water per day.

The various systems of American filters differ chiefly in their details. Let us note the systems of Jewell, Howatson, the New York company, Bell, Reeves, Krenke, and Zimin.

Jewell filters, both gravity (Figs. 63, 65) and pressure (Fig. 64), are sometimes arranged with a settling reservoir under the filter (Figs. 63, 64)—for small installations—and sometimes without it (Fig. 65)—for large stations, where it is more advantageous to arrange separate settling basins. In the first case, the filter consists of two reservoirs placed one inside the other; the inner reservoir serves as the filter, and the space between the walls and bottoms of the two reservoirs serves as the settling tank. Water, with a coagulant solution added to it, enters the space between the walls of the reservoirs; as it passes between them, it leaves a flocculent precipitate on the bottom and, rising through a pipe entering the inner reservoir, pours out over the surface of the filtering sand. The filtered water is collected by drainage tubes laid on the bottom of the filter; along each tube a row of perforated caps is installed, through which the filtrate seeps. A detail of the Jewell drainage tubes is shown in Fig. 66. From the drainage the water is led off to a Weston regulator.

Jewell filters are made with a diameter of 6–24 feet and a height of up to 14 feet. When Weston regulators are used, a head of 14 ft., measured from the water level on the filter to its level after filtration, is sufficient for filtration.

To wash the filter, water is admitted into the drainage under a pressure of about 6 feet; this is done by closing some valves and opening others. The rakes for stirring the sand during washing of the filter are visible in Figs. 63 and 65; they are set in rotation by means of bevel gears from some motor (usually by belt drive).

Fig. 68.

Fig. 70.

The polluted wash water overflows over the upper edge of the filter reservoir and is discharged into the gutters.

The cost of the initial installation and operation of Jewell filters, as well as of other filter systems, is given below in a comparative table compiled from English data.

The general arrangement of a group of filters is shown in Fig. 125, representing the open Howatson filters at the St. Petersburg filtration-ozonation station. Howatson filters differ from Jewell filters in their simpler drainage arrangement (a perforated double bottom, without drainage pipes) and in the use of crushed quartz instead of natural sand.

Crushed quartz has the advantage over sand that it is very hard and its grains are not spherical, but have sharp edges; thanks to this, it is cleaned much better than natural sand. In addition, quartz absorbs absolutely no water; natural sand, however, usually contains an admixture of particles of various kinds of sandstone, often soft and porous, and with constant filtration of water they absorb organic substances and other impurities. Eventually these particles become so contaminated that it becomes entirely impossible to wash them, and the entire filtering material has to be replaced.

Although crushed quartz is much more expensive than sand, its advantages are so great that it should be preferred to natural sand, especially in the purification of waters containing many organic impurities. Crushed quartz almost never requires renewal, provided that, owing to the design of the apparatus, there are no losses of quartz (washing out) during its washing.

The Reeves filter (Fig. 68) has the shape of a truncated cone, turned with its broad base upward; with this shape the water cannot slide down along the walls without being filtered. The filtering material is crushed quartz, whose grain size gradually increases downward. Instead of drainage pipes, there is a double bottom with openings in the upper bottom. This filter operates either under pressure, or with a large difference in the water levels before and after filtration.

Fig. 69 represents a pressure filter of the New York Company, consisting of a closed iron horizontal cylinder filled with sand. Its principal difference from Jewell filters is the absence of special devices for stirring the sand.

The Bell filter (Fig. 70), operating under high pressure (from 2 to 20 atmospheres) ³), differs from the Jewell filter by more elaborate devices for washing the sand. The wash (fil-

³) Installations at Bury, Crewe, Polton, and Linwood.

Steam inlet during sterilization

Section along E–F

Clean water

Control. Outlet of clean water.

Plan along G–H

Introduction of coagulant.

Inlet of dirty water.

Wash-water [[unclear: label]]

Fig. 72.

Fig. 75.

Fig. 71.

...filtered) water is directed not only from below upward through the drainage, but also through agitators made of hollow tubes with perforated walls; the axis of rotation of the agitators is also hollow, and through it water is supplied, which is sprayed through the openings of the tubes in thin jets. In the drawing (Fig. 170) the following designations are used: A—valves on the hollow rods (agitators); B—teeth on the agitators; C—hollow rods (agitators) containing water for washing; D—hollow axis of rotation of the agitators; E—perforated vessels; F—water-discharge pipe for untreated water; G—outlet pipe for cleaning the filter; H—main pipe; J—vertical washing valve; K—outlet of filtered water; L—iron casing of the filter; M—main branch pipe fastened to the casing; N—inlet valve for untreated water; O—outlet valve.

Among the mechanical filters constructed in Europe, we shall note the Krenke and N. P. Zimin filters.

The Krenke filter (Fig. 71) is a horizontal cylinder with double bottoms, the inner bottoms being made of a fine copper mesh; by means of similar meshes the middle part of the cylinder (along its length) is separated off. The spaces between the mesh bottom and the inner mesh nearest to it are filled with filtering sand; raw water enters through the hollow axis of the cylinder, resting on a bearing and connected (by a stuffing box) with a fixed supply pipe, and percolates through the sand in a horizontal direction (from the ends toward the middle of the cylinder). The middle chamber serves to collect the clean water, which flows out of the filter likewise through a pipe coinciding with the axis of the cylinder, but on its other side than the supply pipe. Cleaning is carried out by a reverse jet of clean water under pressure, while the cylinder rotates about its axis, which causes the sand to move and mixes it with the washing water.

The vertical arrangement of the filtering surfaces adopted by Krenke, while increasing the useful area of the filter, greatly worsens the results of filtration, since with it there can be no uniform formation of a filtering film from the settled coagulant. Therefore the Krenke filter can be used only for removing coarse suspended matter, but not for the bacterial purification of water.

The pressure filter of N. P. Zimin (Fig. 72) consists of a vertical iron cylinder filled with sand, in which, around the circumference, a row of tubes nn is installed, supplying raw water, and a central tube C (of large diameter), collecting the purified water. Raw water with coagulant added to it enters the tubes nn from the annular box B, which serves as the base of the filter. Clean water is discharged from the central tube C at its top. All the tubes n and C are iron, with perforated walls, covered with a fine copper mesh.

When washing the filter, besides the reverse jet of water, one may use a jet of steam, introduced through a special tube at the top of the cylinder.

A disadvantage of Zimin’s filter, as of Kröncke’s, is the vertical arrangement of the filtering surfaces.

To reduce the area occupied by American filters, they are placed one above another in several tiers.

As examples of large installations for water purification with American filters, we shall mention the station in Petersburg for purifying 50,000 cubic meters of water per day, equipped with Howatson filters, with subsequent sterilization of the water by ozone (see the description below), built in 1910, and the station in Cincinnati, operating since 1907.^4

The filtration station in Cincinnati was built on the basis of prolonged experiments with American filters and is designed to purify 35 million vedros (= 430,000 cubic meters) per day.

In view of the large size of the station, certain simplifications were made in the construction of the filters.

Water (from the Ohio River) is taken 12 km from the city, upstream of the river, and is delivered through two cast-iron pipes (\(d = 60\) in.) into settling tanks located 60 meters above the river level.

All the valves of the pumping station are piston-type hydraulic valves.

Settling tanks, two in number, with a total capacity of 100 million vedros, are constructed in the form of ponds in the valleys of two small tributaries of the Ohio River, blocked by dams; to achieve impermeability, the bottom and sloping walls of the basins are lined with a layer of clay 3 feet thick, covered with a layer of crushed stone and concrete 0.5 foot thick and a double layer of asphalt. These settling ponds are used in alternation; consequently, settling lasts about 36 hours. Pressure pipes with hydrants are laid around the ponds for washing the ponds by means of hoses.

Coagulation. From the settling ponds the water is taken at a constant depth of 3 feet from the surface^5 and through two pipes (\(d = 60''\)) is directed to the filtration station (Fig. 73); on entering the station the water enters 6 parallel pipes (\(d = 36\) in.), equipped with pressure regulators (since the water level in the settling ponds is 10 m higher than on the filters). In a small chamber a coagulant—ferric sulfate (in the amount of 16–40 g per liter)—is added to the water; its solution flows here from the chemical department through a 3-inch pipe. Mixing is carried out by small propeller pumps. Next the water passes through Venturi meters into the valve room, in which, to it,

^4 “Engineering Record,” 1907 (April 6, April 8), 1908 (February 1), 1909 (January 16). See also the report of N. P. Zimin to the 9th Waterworks Congress.

^5 By raising the intake ends of the pipes to a float.

Fig. 73.

Fig. 74.

lime water is added (delivered through pipes \(d = 24\) dm), in an amount of 12–25 g per liter, and then it enters the open coagulation settling tanks.

Preparation of the coagulant solution. In the chemical department, lime is slaked in iron tanks (\(d = 6'\) and \(h = 3'\)) with mechanical stirrers; three tanks are used simultaneously, while the other three serve as reserves. The milk of lime obtained from slaking is then diluted in 4 tanks, each of 8,000 buckets capacity, until the solution becomes completely transparent; this consumes about 5% of all the water entering the station.

The lime is poured into the dissolving tanks through hatches in the floor of the upper story.

Ferrous sulfate is dissolved in filtered water in 5 iron tanks, each 5 ft in diameter and 12 ft deep.

The quantity of reagents needed for coagulation is assigned according to the quality of the water; each time 35 kg are poured into the mixer, but the intervals of time between additions are determined by the head of the laboratory, and special clocks give signal bells at the appointed equal intervals of time. In this way, no weighing of reagents is required, and they are dissolved by two workers, who add one sack at each ringing of the clock.

The settling tanks for coagulated water are arranged as 3: two with a capacity of 3 million buckets each and one of 1/2 million; by connecting these settling tanks, a settling duration of from 2 to 4 1/2 hours can be attained. The flow of water through the settling tanks is directed by partitions. The bottom of the settling tanks has a slope; the sediments are removed through drainage pipes. If necessary, an additional quantity of coagulants may be introduced on the way from the settling tanks to the filters.

Rapid filters are arranged in a number of 28, each for 1 1/2 million buckets per day, with a surface of \(32' \times 50' = 1600\) sq. ft. The filtration rate is 170 dm = 430 m per hour. For ease of repair, each filter is divided by a wall into two independent halves.

The filters are arranged in two rows on both sides of the gallery of distributing pipes, in a covered heated room. (Fig. 74). The entire construction is reinforced concrete.

The thickness of the filtering layer of sand is 30 dm = 76 cm; the effective size of the sand is 0.35 mm, coefficient of uniformity 1.60.

The drainage consists of vertical tubes entering into a trough-shaped concrete bottom (Fig. 75), the depressions of which are filled with gravel of size from \(1/12\) dm to 1 dm; on top of the gravel is placed a wire brass mesh with openings of \(1/10\) dm, which retains the gravel during washing; this mesh is fastened into the concrete with bolts.

All the valves of the filtration station are equipped with electric motors and are controlled from a distribution board. Open or closed po—

the position of the sluice gates is indicated on a board by small lamps. The wash-water sluice gates can be opened by electric current to any height.

The filtration-rate regulators (Viviani system) are based on the principle of a reducing valve: the difference in pressure on the two sides of the hydraulic sluice gate through which the water flows is automatically kept constant; this pressure difference, acting upon an elastic diaphragm, controls a cock that admits water to one or the other side of the piston of the hydraulic sluice gate, and thereby correspondingly increases or decreases the opening closed by this piston.

Washing of the filters is carried out (on the basis of preliminary experiments) in a simplified manner, without mechanical stirring by rakes or the blowing-in of air, but only by a strong current of water from below upward. The velocity of the wash water is taken as 2 feet per minute, i.e. twice as great as is usual in American filters with mechanical stirring of the sand. The consumption of wash water has not thereby increased, since it proves sufficient to admit the wash water at double velocity for only 3–4 minutes. The filter has to be washed after 20–30 hours; the filter is taken out of service for 10–15 minutes.

At the beginning of washing the wash water is admitted gently; the polluted filtering film is thereby lifted bodily upward and removed through overflow troughs located 30 dm above the surface of the filtering sand. When in this way the greater part of the substances polluting the filter has been removed without mixing with the sand, the wash water is admitted in a strong stream. The sand is stirred up by the water, lifted by it, and kept in an elevated position, so that its surface is 15 dm above normal. After washing ceases, the sand settles, sorting itself well: the coarse particles lie below, the fine ones above, which is very favorable for filtration.

The wash water enters the filter from below through the drainage (under a pressure of 1/2 atmosphere) from a special reservoir situated on a neighboring hill, with a capacity of 60,000 vedros (a reserve sufficient for washing two filters for 4 minutes), from which the water is led by a pipe (\(d = 36''\), about 120 m long). To deliver filtered water into this reservoir, two centrifugal electric pumps are installed, each of 45,000 vedros per hour.

These pumps are set in motion and stopped by switching the electric current on and off by means of a float in the reservoir, so that the reservoir is filled automatically.

Reservoir of purified water. The filtered water from each filter passes through the regulators by 30″ pipes into a collecting 60″ pipe and flows by gravity into the clean-water reservoir, located

next to the filtration station; the capacity of the reservoir is 6 million buckets, the area \(400 \times 400 = 160000\) sq. ft. The reservoir was made open, but the foundation is arranged so that it would be possible to cover the reservoir with vaults.

Supplying water to the city. Water from the reservoir flows by gravity through a brick tunnel (about 7 kilometers long) to a pumping station located in the city and forcing the water into the network.

The cost of constructing the filtration station (not including the cost of the large settling ponds and pumping stations) is about \(2 \tfrac{1}{2}\) million rubles, i.e. 7 kopecks per daily bucket (or 5 rubles 60 kopecks per 1 cubic meter). Of this sum, in round numbers, the filters together with the building account for 1 million rubles, the cast-iron pipelines for 0.5 million rubles, the coagulation settling basins with devices for 0.5 million rubles, the clean-water reservoir for 0.25 million rubles, and the laboratory together with the chemical department and their equipment for 0.25 million rubles.

The filtration of the water during 1908 cost 130000 rubles, i.e.

\[ \frac{130000 \times 100 \times 100}{35000000 \times 360} = 0{,}103 \]

kopecks per 100 buckets.

Results of water purification by American filters.

In order to determine the action of American filters, the following questions are especially important:

1) Does the addition of a coagulant to the water impair its quality.
2) Does the artificial film formed on the filter as a result of the presence of coagulants possess the same properties as the film of English filters.
3) What are the results of purification from the bacteriological standpoint.
4) Can and should every water be purified on American filters.

The addition to the water being purified of coagulants, and in particular the most widespread of them—aluminum sulfate, is apparently entirely immaterial from the hygienic standpoint, provided only that the reagent is chemically pure; aluminum compounds also occur in natural waters. Moreover, artificially added alumina is completely decomposed during coagulation, and the small amount of gypsum that appears in the water in its place (increasing the permanent hardness of the water) presents no inconvenience from the hygienic standpoint. It is only necessary to avoid adding aluminum sulfate in such a quantity that, if there is an insufficient content of carbonic lime in the water, free sulfuric acid could be formed.

Furthermore, the aluminum sulfate available commercially often contains about 1% arsenic; however, the greater part of this arsenic precipitates in the settling reservoir in the form of insoluble arsenic-

aluminum sulfate or arsenious-acid alumina. Objections to the enrichment of water with arsenious compounds should rather be regarded as theoretical and as having no practical significance, since up to now in America not a single case is known of impairment of health from the use of water purified by the addition of coagulants.

Many investigators have dealt with the question of whether the artificial film of American filters is as reliable a means of retaining bacteria as the film formed naturally on English filters. Bitter and Gottschlich⁴ found that, 30 minutes after American filters had begun operating, they possessed such retaining capacity that the filtrate could be used as drinking water without any danger. Specific bacteria at that time passed through in the amount of 1 out of 10,000 present in the raw water; from a certain moment (1–2 hours after the filter began operating) 1 out of 15,000 bacteria passed through, whereas during the first half-hour one out of 2,000–3,000 bacteria passed through.

The number of ordinary bacteria harmless in water, after half an hour had elapsed, was in all experiments below 100 in 1 cubic cm of filtrate, and in most of them below 50. These bacteria, for the most part, did not belong to the number of bacteria of the raw water (i.e., they had entered the water from the filter).

It should be noted that, in the process of settling with coagulants, the number of bacteria in the raw water decreased by 15–75%.

According to Schreiber’s investigations¹⁷ on Jewell filters, the filtration rate can be considerably increased above the generally accepted value of 5 meters per hour,⁵ while their ability to retain bacteria remains equal to that of slow sand filters. The reduction he observed in the number of bacteria in the raw water amounted to 91.3–99.6%.

With respect to the removal of turbidity and color, American filters are incomparably superior to English ones; precisely the circumstance that in America it is often necessary to use, for water supply, very turbid water colored green, as we have mentioned, led to the spread in America of these filters chiefly in preference to English ones.

However, good results cannot be achieved with every water on American filters without additional treatment. First of all, with very soft water, as we have already said, there can be no formation—

the formation of coagulant flocs, and the water may acquire an undesirable content of undecomposed coagulant. To avoid this, lime has to be added to soft waters, and sometimes a small quantity of soda is also added.

Secondly, water with a high content of fine silty particles, and especially of plankton (the latter is sometimes found in reservoirs with stagnant water), requires the addition of a large quantity of coagulant; in this case the filter quickly becomes clogged, the required pressure on the filter rises rapidly, the quality of the filtrate deteriorates, and the filter has to be cleaned often. Therefore, when there is a high content of silt and plankton, it is more advantageous and better to remove part of the impurities from the water before filtration—for example, by passing the water at the intake through fine screens or even through a layer of fabric (stretched on a frame instead of a screen).

Finally, for water whose character and degree of turbidity vary strongly and constantly during the year, the use of American filters would require constant changes in the quantity of coagulants added, i.e., constant analyses of the water and constant careful supervision of the filtration process would be required. Therefore, generally speaking, it is preferable to filter such water on English slow filters, or else, in the case of a high content of turbidity in it, to combine American and English filters, using the former as a preliminary treatment of the water (to clarify it) before its final purification on English filters, which, with clarified water, yield a filtrate of more or less constant quality⁶.

Thus, American filters, being suitable for purifying almost all waters (with an appropriate choice and quantity of coagulant), prove to be specially applicable, in preference to all other systems, for waters with a constantly large quantity of turbidity; when the turbidity content is constant, operation of American filters is extremely simple, and a large quantity of turbidity can be well removed only by these filters.

The cost of constructing American filters, generally speaking, is less than that of English filters. Owing to their considerable filtration rate, they occupy a very small area, which, for economic reasons, compels giving them special preference in places where land is expensive, and also makes them convenient for installation in separate houses; in the latter case closed filters operating under pressure are more often used (utilizing the pressure of the municipal water-supply network).

TABLE A

Comparative Cost of American Filter Systems

Referred to a water consumption of 1,000 cubic meters per day.

Prices are according to English data; 1 pound sterling is taken as 9 rubles 60 kopecks. The cost of the reservoir for clean water is not included.

	Jewell	Bell	Reeves	Candy
Filtering area	9.3 sq. m	6.2	20	6.5
Cost of installation	9,500 rubles	4,650	5,300	4,700
Daily operating cost, not including interest	1.27 rubles	1.06	Not fully established. Approximately the same as for the Bell filter.	0.21
How often the filter is washed	1 or 2 times per day	From 1 time per day to 1 time in 3 days.	Usually 1 time per day.	From 1 time per day to 1 time in 3 days.
Duration of production of turbid water after washing the filter	15 minutes	When aluminum sulfate is added, several minutes.	3–5 minutes.	Several minutes.
Quantity of wash water, as a percentage of the quantity filtered	\(2 \tfrac{1}{2}\)–5%	1% or less.	About 2%.	\(1/2\)%.
Service life of the filter	50 years	Equal to the service life of iron bridges.	Not less than 15 years.	50 years.
Devices for regulating the quantity of added reagents (coagulants)	Special regulating openings are provided.	Automatic regulators are attached.	Regulator with a ball valve, or a small pump with a water engine.	Not required.

The cost of water purification on American filters, owing to the use of coagulants, is somewhat higher than on English ones, although close to it.

Let us present some figures for the comparative cost of operation, referred to 1 million gallons of water, in shillings⁷.

Authority	City	Cost of ordinary filtration	Cost of mechanical filtration
Fuller	Cincinnati	10,35	9,96
Hazen	Pittsburg	10,53	10,78
Miller	Washington	8,50	8,76

The appended table gives comparative data on filters of various systems (according to English sources).

Thus, the chief obstacle to the spread of American filters is their sensitivity (in particular, the sensitivity of the coagulation process) to changes in the composition of the water, and partly also the lesser confidence of European researchers¹⁸ in the assurance of the bacterial purity of the filtrate, which, however, finds no grounds in the many years of experience of American cities.

Chapter VI

Filters of Special Systems

In all the filters we have considered, sand served as the filtering material; but attempts were also made to employ other filtering materials: porous tiles made of artificial sandstone, plates of unglazed porcelain (“biscuit”), charcoal—both wood and animal charcoal—fabrics (cloth, felted woolen fabric, or asbestos fabric), sponges laid in a continuous layer, and, finally, infusorial earth.

However, by no means did all these materials prove satisfactory, and many filters found application only in isolated cases. We shall describe the systems that gained some degree of use—the Fischer and Peters filters, Kurk’s filter, and the domestic (room) filters of Chamberland and Berkefeld.

The Fischer and Peters filter (Fig. 76) consists of a series of hollow vertical plates made of artificial sandstone; the hollow space inside the plates communicates with a water-collecting pipe that gathers the filtered water. The plates are installed in a reservoir into which raw water is admitted for filtration; this water seeps through the pores of the plates into their internal cavities and is then led from there through the water-collecting pipe into the clear-water reservoir. The plates are washed by passing clean water in reverse from the collecting pipe outward through the plates. After prolonged operation of the filter, when the contamination has penetrated deeply into the thickness of the plates (the plates “grow through”), the plates are removed and calcined.

The filtering surfaces were placed vertically in order to reduce the area of ground occupied by a filter having a considerable filtering surface; however, as we have already mentioned in speaking of Krecke’s filters, the seepage of water through vertical surfaces proceeds very unevenly at different depths, and the formation of the filtering film is not only uneven but also very weak.

At first glance one might have expected that the fine pores of artificial sandstone would retain impurities and bacteria better than the pores of a layer of sand; however, experience has shown that plate filters have no advantages over sand filters in the quality of purification and thus possess only a major drawback—the unevenness of operation and of the filtrate quality. Washing them with wa-

Fig. 76a.

Fig. 76b.

proves to be very imperfect and poorly amenable to control, while firing the tiles is very difficult; the tiles break easily; moreover, an admixture of turbidity in the water leads to rapid clogging of the tiles.

In view of the foregoing, the Fischer and Peters filters, as well as artificial-stone filters in general, are at present no longer used in the central water supplies of cities.

The Kurke system differs from the preceding one in that the artificial porous stone is made in the form of cylinders 1.20 m long and 0.22 m in diameter, joined (by square bases) into batteries (of 16 cylinders) measuring 1.00 × 1.00 meter in plan. Filtration takes place (from outside into the cylinder) exactly as in the Fischer filters.

Of the materials used for making household filters, only two are capable of retaining bacteria: unglazed porcelain and infusorial earth; the others merely retain turbidity (clarify the water), although, of course, together with the turbidity some of the bacteria adhering to it are also retained.

The use of household filters for the purpose of reducing the bacterial content of water cannot be recommended, since even the best filters can work without cleaning only for a certain time, and, with the variable and intermittent water consumption under which household filters operate, it is rather difficult to establish the intervals between cleanings. Moreover, the usual ignorance or negligence of the persons using these filters makes their operation very unreliable. Finally, the capacity of those filters which (at least at the beginning of operation) are capable of retaining bacteria is very small.

We shall dwell only on the Chamberland—Pasteur porcelain candles and on the Berkefeld filters with infusorial earth.

The Chamberland filter (Chamberland) (Fig. 77) consists of a cylindrical metal reservoir screwed (by its upper end) to a water tap; a filtering hollow “candle” of unglazed porcelain (20 cm long and 2.5 cm in diameter), provided with an opening at the bottom and projecting from the metal casing of the filter, is inserted into the reservoir. Water seeps into the candle and comes out purified from the lower opening.

All investigators recognize satisfactory retention of bacteria from the water by the candle during the first 4—12 days of operation; therefore cleaning and sterilization of the candle is desirable, on average, once a week.

Chemical disinfectants are not capable of sterilizing the candle, and it has to be fired in dry air (in special Pasteur ovens or in ordinary bread ovens) at a temperature of 280°—300° for half an hour.

Fig. 77.

Fig. 78.

The output of Chamberland filters is very low and, in new filters, is equal to about 30 liters per day; after each cleaning it decreases and eventually reaches 3 liters.

Chamberland candles are sometimes connected into a battery of several candles placed in a common reservoir of raw water and having a common water-collecting channel for clean water.

In the Berkefeld filter (Fig. 78) the filtering candle B is made of infusorial earth; water enters (through A) at the top of the cylinder, seeps into the interior of the candle, and exits upward through tube E. To clean the filter, a cylinder N with brushes touching the surface of the candle has been added; this cylinder is set in rotation by handle M; the soiled wash water is discharged through tap V at the bottom of the filter. In addition to cleaning with brushes, it is necessary fairly often to sterilize the candle in water brought to boiling.

Experiments have shown that, despite the large size of the pores of the Berkefeld candle, it does not allow bacteria to pass, but only during the first 3–5 days of operation; therefore sterilization of the filter is required at least twice a week.

The output of the Berkefeld filter is about 0.5 liter per minute, i.e., considerably higher than that of Chamberland.

Let us now turn to filters whose material is intended not only to mechanically purify the water passing through, but also to promote its purification by oxidation.

Houston proposed replacing the crushed quartz in his American filter with a special composition, “polarite,” consisting of 54% iron peroxide, 25% silica, 2% calcium, 6% aluminum, 7% magnesia, and 6% alkalis. The action of this porous composition is similar to that of spongy platinum: it absorbs and condenses on its surface (adsorbs) the oxygen of the air and gives it to the organic impurities of the water being filtered. When, after operation of the filter (for about 6 weeks), the polarite (i.e., the reserve of oxygen in it) “is exhausted,” it is brought into contact with air for 3–4 days, and then it is again ready for operation. The thickness of the polarite layer is about 0.40 m; beneath it is placed a layer of coarse quartz sand 0.10 m thick. The filtration rate is about 6 m per day, i.e., very low.

The Candy filter, which is spreading very rapidly in England, has considerably greater output.

Candy filters are constructed both as pressure filters (Fig. 79) and as gravity filters. As it enters the filter, the water is sprayed in fine streams; to improve aeration during spraying in pressure filters, the upper chamber of the filter contains compressed air. The water seeps through a layer of coarse crushed quartz, tol-

1. Commercial sulfuric-alumina salt usually contains about 16% alumina \(\left(\mathrm{Al_2O_3}\right)\), 35% sulfuric anhydride \(\left(\mathrm{SO_3}\right)\), and 49% water, i.e., about 51% aluminum sulfate \(\left(\mathrm{Al_2(SO_4)_3}\right)\). ↩↩↩↩↩↩

2. Device for feeding a solution of chloride of lime for the disinfection of water. ↩↩↩↩↩

3. With the indicated dimensions, the total area of the forfilters and filters per 1000 cu. m of water purified per day is equal to ↩↩↩↩

4. Bitter und Gottschlich. Zeitschrift für Hygiene 1908, No. 59, 379. ↩↩↩↩

5. In America the filtration rate is sometimes brought up to 8 m per hour. ↩↩↩

6. Friedberger, “Zeitschrift für Hygiene,” 1908, p. 379. ↩↩↩

7. Quoted from Cherepashinsky, Water Supply, p. 167. ↩↩↩

8. According to the investigations of Zigmondy, Burton, Dunbar, and others. ↩↩

9. According to the proposal of the Board of Health of Massachusetts (1892). The effective size of sand is also sometimes called the “productive size” of the sand. ↩↩

10. These dimensions are given as the most advantageous in Allen Hazen’s work: The Filtration of Public Water Supplies. ↩↩

11. On average about 20. ↩↩

12. This interval \(t\) includes the time necessary for drawing off the water, removing the upper contaminated layer of sand, admitting water, and forming the filtering film (the ripening of the filter). On average, \(t\) is approximately equal to 2–4 days. ↩↩

13. In the case of a great demand for water not satisfied by the remaining operating filters. (N. K.). ↩↩↩

14. And not caused by defects in the construction or operation of the filters. ↩↩

15. Piefke. Die Prinzipien der Reinwassergewinnung vermittelst Filtration. Berlin, 1887. ↩↩

16. Köhler. Physikalische Eigenschaften des Sandes. (Dissertation). Karlsruhe 1906. ↩↩

17. Schreiber. Mitteilungen aus der Königlichen Prüfungsanstalt für Wasserversorgung und Abwässerbeseitigung, 1906, VI, 88. ↩↩↩↩↩

18. Friedberger—the above-cited article. Hilgermann.—“Vierteljahresschrift für Medizin und öffentliches Sanitätswesen,” 1906, 36. ↩↩

19. Allen Hazen. This formula is used to determine the loss of head \(h\). ↩

20. In bacteriological analysis one should pay attention to whether the bacteria found at different times in the filtrate belong to one and the same species or to different ones; in the first case the bacteria enter the water from the sandy layer of the filter, while in the second they were contained in the raw water. ↩

21. Reinsch, “Zentralblatt für Bacteriologie” 1894. ↩

22. This instruction (“Grundsätze für die Reinigung von Oberflächenwasser durch Sandfiltration”) was originally issued in 1893, and then, in a revised edition, in 1898. We cite only some of its paragraphs. ↩

23. “Brief Description of the Moscow Municipal Waterworks. 1910.” ↩

24. “Notice sur les Procédés d’épuration des eaux de rivière, employés par la Ville de Paris à l’Usine d’Ivry”. Paris, 1908. ↩

25. To reduce such seepage along the wall, the surface of the side walls of the filter is sometimes made rough. ↩

26. Not shown in the drawings. ↩

27. The cost of cleaning 1 sq. m of filtering surface in Warsaw is 95 kopecks. ↩

28. In Berlin—30 cm, but this thickness is scarcely sufficient; it is desirable to limit the minimum to 40–50 cm or more. ↩

29. According to Debauve et Imbeaux (“Assainissement des Villes. Distribution des eaux.” I, 595). ↩