Robert E. Milligan.

While much of the information contained in this paper is admittedly an old story, especially to the Water Analyst and Bacteriologists, it will be useful to reiterate the facts for the benefit of the ever-increasing number of filter operators who, as a matter of fact, have little or no opportunity of gathering from various sources the information herewith presented. The mechanical filter does not depend upon its sand and gravel filtering bed alone for the high efficiency obtained. Coagulation and sedimentation are also essential features of the process and possibly least understood by the operator. Coagulation is produced by the introduction into the raw water of a soluble, harmless, mechanical salt, capable of decomposing when brought into contact with alkaline salts into an insoluble hydrate technically known as “Coaguluum.” There are three soluble salts used; i. e., Sulphate of Alumina, Alum and Sulphate of Iron.

The following reactions occur when any of these come into contact with alkalinity which may be native in the water or may be supplied through introduction of alkaline salts; commonly Soda Ash or Hydrated Lime.

SULPHATE OF ALUMINA REACTION. (AL2SO4) + 3 CA (HCO3)2 =3 CASO4 + 6 CO2 Hydrate + AL2 (OH)6


K2A12 (SO4) + 3Ca (H Co3)2 = 3 CA SO4+6 Hydrate

CO2 + Al2 (OH)6 + K2 SO4

It will be noted that the result is substantially the same where Alum or Sulphate of Alumina is used as a coagulant.


2 Fe SO4 + 2 Ca (OH)2 + H2 O + O = 2 CaSO4

Hydrate + 2 Fe (OH)3

Sulphate of Iron does not react satisfactorily with native alkalinities, therefore it is necessary to apply Hydrated Lime. Sulphate of Alumina containing 18 per cent. Alumina Oxide requires 9 p. p. m. Calcium bicarbonate (Ca (H CO3)2) to complete react through a filter plant; and on an average 3.5 p. p. m. of Carbonic Acid (CO2) is set free by each applied grain of Sulphate of Alumina when decomposed through reaction with the bicarbonate. The theoretical amount of Co, set free is 6.77 p. p. m. for each grain of alum. This free Carbonic Acid, it is suggested by some experts, tends to loosen and render active certain organic acids, common to highly colored waters and the general effect is to assist in the destruction of the color common to many of the raw waters needing purification. It is generally stated, however, that the color in the raw water is absorbed by tbe “Floe” or Hydrate of Alumina, known as the “Coaguluum.” The action of free Carbonic Acid on exposed iron surfaces and the possibility of it producing or increasing the condition known as “Red Water,” is still under discussion. This condition which has become a nuisance in many instances is attributed to one of three causes, that is to say, the Carbonic Acid Theory, the Electrolytic Theory and the Hydrogen-Peroxide Theory. Some experts are inclined to the Electrolytic Theory as the cause of the corrosion, but there is a great deal of data that indicates the correctness of the acid theory and whether it is directly the action of Carbonic Acid Gas or of Dissolved Oxygen is not clearly established. In any case, from the standpoint of the filter operator, the passing of acid water or acid filtrate into the mains and service pipes is to be avoided.

No supply should be passed through the mains and service pipes where alkalinity is less than 5 p.p.m. and it is a sure remedy for the “Red Water” condition and the prevention of Crenothrix growths, if the lime be applied in slight excess. The lime insulates the exposed iron surface, particularly wrought iron, against the possible effect of acid water and saturates the free CO,; and the presence of at least 5 p.p.m. of alkalinity in the filtrate insures the complete reaction of the Aluminum Sulphate to the hydrate form and prevents “After-Coagulation” in the filtered water. It is of the utmost importance that the coagulation be applied under regulated control, through devices that measure the exact dosage to the water passing to the filtered plant and in ratio to the volume treated. The Coagulant is handled in solution ranging from 2 1/2 to 5 per cent, in strength and is measured continuously and proportionately into the applied water. Where there is little or no fluctuation in the pumpage or flow to the filtration plant, calibrated orifice tanks are commonly used. Where variations occur in the volume of the applied water passing to the filter plant,—frequently devices of the proportionate feed type are employed. These are of several varieties, and descriptions of them, which would occupy too much space on this paper, can be obtained from the Filter Companies. Dry feeds for applying coagulatns (where a ground or powdered form of the chemical is employed) are also used. These dry feeds also are proportionate and a dry powder is applied under control and regulation. This last method has not shown any particular advantage over the volumemetricmethod as yet; except in the case of lime feeds. The hydrated lime does not enter freely into solution and it is not economical to store the lime solution; the strength of which is from 50 to 70 grains to the gallon. The use of milk of lime has many drawbacks, though this concentrated solution is and can be used through special devices designed for that specific purpose.

In coagulating water the result to be obtained is the formation in the raw water of a fully formed, insoluble harmless compound (Hydrate), which, while aggregating together to a mass condition, is capable of absorbing within the enveloping coaguluum practically all the suspended impurities and color with which it collides. The water is then coagulated and can be successfully purified by sedimentation and filtration.

It is worth remembering that all bacteria, both pathogenic and non-pathogenic, arc in suspension and are removed by the filtration process; and that some forms resist sterilization, even in cases where prohibitive quantities of chlorine gas are used, so that the removal of bacteria in large percentage amount is essential as a preliminary to sterilization. The above reactions and descriptions of coagulation includes under the term “coagulants,”—applied alkalies, such as lime and soda, as well as sulphate of alumina and sulphate-of-iron; so far as the purpose of this paper is concerned.

Most surface and well waters contain sufficient alkali in solution to react completely with the small amounts of applied sulphate of alumina necessary to remove impurities, and will leave a sufficient alkaline filtrate to prevent any “aftercoagulation” in the filtered water, and in such cases applied lime would be used only to prevent conditions as described above in connection with “Red Water.” There are many cases of fluctua tions in alkalinity and there are cases where an insufficient amount of alkalinity exists. There are also raw waters that are acid to reaction and it is of the first importance that the operator should determine the alkalinity or lack of it in the raw water and this is accomplished by alkalinity tests. The method generally employed is known as the “Erythrosin Method” and requires the following re-agents:

First: Solution of 50th Normal Sulphuric Acid.

{N/50 H2SO4} 1 which is prepared by standardization against Sodium Carbonate 50th normal solution

{N / 50 NA CO3} The latter solution is prepared

by weighing out 1.06 Grammes of pure Anhydrous Sodium Carbonate (made by heating the carbonate and bringing it up to one Litre in volume (1,000 grammes) with distilled water).

Second: Erythrosin Solution—Dissolves 250 Milligrams Erythrosin in one litre of distilled water.

Third: Cholorform.

The process follows: Use 250 C. C. white glass flat stoppered bottle. Int6 the bottle place 100 C. C. of the water to be examined. Then introduce one C. C. of the Erythrosin Solution ; add 5 C. C. of the chloroform and thoroughly shake the mixture.

Then from a burette or graduated pipette add drop by drop the 50th normal sulphuric acid solution, shaking the mixture after the introduction of each few drops. It will be noted that the rose color of the mixture becomes fainter as the acid is added, until sufficient is used to completely discharge the color, when the “end” point is reached and the test completed. Then read on the burette, the cubic centimeters and fractions thereof used to discharge the color. The amount of acid so read, gives the alkalinity of the waters in parts per hundred thousand; multiply by ten in order to express in parts per million; for example: If 1 and 3/10 C. C. of acid be required to discharge the rose color and reach the “end” point, the alkalinity of the water is 1.3 p. p. per 100,000 or 13 p. p. m. in terms of Calcium Carbonate. Do not add acid beyond the “end” point and compare the discharge of color with the blank sample of the water being tested. For the convenience of those operating small plants without laboratories attached or for field work, convenient portable alkalinity sets can be purchased. The Erythrosin test is used on the assumption that the raw water is alkaline and not acid. This condition is readily observed in the color of the mixture before adding any acid, if alkaline the mixture will be rose tinted; if acid it will remain colorless and if the mixture be neutral it will also show an absence of the rose tint.

It is essential to test the raw water at the same time the filtrate is tested in order to make comparison and often the raw water is muddy with turbidity or extremely high in color. It is, therefore, necessary to prepare the raw sample for test in such cases so that the “end” point can he read accurately. In many cases the turbidity can be removed sufficiently by filtering the raw sample through filter paper; in some cases, however, it may be necessary to prepare a paper pulp made from white filter paper, disintegrated with distilled water. By shaking this up thoroughly with the raw sample and then filtering through the filter paper, the turbidity can be entirely removed. In extreme cases, color may be removed by percolation or slow filtration through freshly prepared hydrate of alumina, rendered neutral by repeated washing with distilled water until the last rinsings are neither acid or alkaline to test, A comparison of the difference between the alkalinity of the raw and filtered sample, establishes a rough measure of the amount of sulphate of alumina actually decomposed in the process of coagulation, since it requires approximately 8 parts of alkalinity per million to decompose one grain of 17 1/2 per cent, sulphate of alumina to the full formation of the coaguluum. (Hydrate of Alumina). The Hydrate as it forms and becomes the “floe” aggregates together and envelopes the particles of suspended matter in the raw water and absorbs color. The “floe” increases in spccific gravity becomes heavier than the water and precipitates or subsides towards the bottom. This explains in large part the value of the subsidence, or settling basins and their necessity as part of a mechanical filtration plant handling turbid waters. The basins also provide means of thoroughly mixing the coagulant with the raw water, and furnishes the time necessary with some water to complete the reaction and form the Hydrate “floe.” The settling portion of the filter plant should remove at least 60 per cent, of the suspended matter from the water passing to the filters. Color reduction is usually not so great in the basins, though the filtrate will show a color removal equal to 90-to-95 per cent, as compared to the color in the raw water. The basins should be washed out at frequent intervals, not more than a foot of liquid sludge being allowed to accumulate in the bottom. It is advisable that the operator should know the turbidity of the water, as expressed by the turbidity scale. Surface waters range from 0. to 8,000 parts per million of turbidity and great fluctuations occur, commonly rivers range from 20-to-500 p. p m. There is no accurate relation, yet established, between the p. p. m. of turbidity and the grains per gallon of alumina required to coagulate the turbid water. A rough relation does exist nevertheless and applied sulphate of alumina should be increased or decreased as the turbidity increases or decreases. The operator soon becomes skilled in determining this, having before him both the raw and filtered samples for comparison. Practically all the methods used or offered are described in “Water Supply and Irrigation Paper No. 151” Series L. Quality of water 11.” Department of the Interior U. S. Geological Survey—Washington, D. C. In my opinion the most practical method is to prepare artificial standards in jars covering the range of turbidity of the particular water being tested. The standards may be prepared as follows: Fullers Earth (Diatomaceous Earth) is used to represent the finely divided clay causing turbidity in the water. (Pears precipitated Fullers Earth can be obtained from Druggists and is satisfactory). One gramme, 15.5 grains is weighed out and placed in one (1) litre of distilled water. The mixture represents 1,000 parts per million of Silica Turbidity standard. Dilutions from this can be made to suit the varying turbidity of the water it is intended to be a counterpart of. Sample of the turbid raw water is matched against the standards (after shaking) until a close comparison is established, then from the standard may be directly read the p. p. m. of Turbidity. Roth the standard and sample should be in front of light when read. The disappearance of a light due to the opacity of the turbid water is cleverly worked out by the Jackson turbidity method. It is described in the above-mentioned government paper and the apparatus called the Turbideter can be obtained from Eimer and Amend, New York, costing less than $20; it is made in two styles, candle, or electric light effect. The water is poured into a graduated standard glass tube enclosed in a dark rase, and superimposed over the light the rays from which penetrate upward through a cross cut out of the dark cover or case enclosing the standard tube. The operator continues to pour water gradually into the tube until the image of the candle flame just disappears. The standard tube is then removed and the turbidity is directly read from the subdivisions marked on the tube:

The writer often prepared a standard by allowing the native turbidity of the raw water to subside naturally and using this silt thoroughly dried to prepare the several comparison solutions in waters heavy with turbidity, a very close reading is thus established. As the coagulated water passes to the Filter units the size and appearance of the “Floe” should be observed. The inflowing water should carry upon the Filter bed a sufficient “Floe” to ensure the mehranous layer or “schmutzdecke” upon and within a foot of the sand bed surface. This layer of flocculant gelatinous Hydrate is the real filtering media and interposes itself between the following inflowing coagulated water and the sand bed the interstices or voids between which arc themselves sufficient in diameter to permit the passage of Bacteria and in many cases silt, i. e., turbidity to pass through the bed into the strainer system and thence to the consumer. As this laver of coagulated impurities is removed from the water and deposited on the filters, it increases in depth and consistency. The frictional loss or resistance of the filter bed increases and this is indicated and often recorded by a loss of head gauge. This device gives the difference in frictional loss between the water table on the filter and the filtered water as it leaves the controller to the clear well in feet and fractions thereof—a full description of these registering or indicating devices can be had on application. As the filter becomes dirty the flow through it decreases until it is economical and expedient to wash it. In washing a filter unit it is necessary to “float” the sand bed without disturbing the gravel layer, usually 9 inches in depth, placed above the strainer system and upon which the sand bed rests. Except in instances where a high velocity wash is employed the filter is washed at the rate of 8 gallons per minute per square foot of sand bed surface (upward flow) and with high velocity wash a rate of not less than 24 inches equal to 14.96 gallons per square foot of surface is necessary to float and cleanse the sand bed. The washing methods employing the low rate of wash water (8 gallons) depend upon mechanical help as well as upon the upward flow of the water itself and agitators—air under pressure and sectional washing devices are employed to assist the cleansing process. High velocity methods have failed until recently because of the unseating of the gravel layer which becomes interspersed among the sand grains and lessens the efficiency of the sand bed. Within the last 18 months, however, the “Wheeler Bottom” has come into use and a very high velocity wash is maintained without disturbance of the gravel layer. Carefully conducted and observed tests show that as high as 48-inch vertical rise equals 29.92 gallons per square foot per minute of surface area can be utilized without any movement of the gravel. Description of all washing devices may be had on application. The filter should be washed with filtered water until all the adhering impurities are discharged upward from the filter bed and the wash water as it flows through the gutters to the sewer outlet shows a better appearance than the raw applied water until it is practically free of suspended matter and color. Rewash connections are installed on most filters and they enable the operator to filter downward at a rapid rate and waste the first roughly filtered water to the sewer. Thus the bed is drawn back into place and compacted. A certain amount of coagulated matter coats the surface and the filter is ready for efficient work at the normal rate of filtration. It is good practice to construct filters over the clear water well and enclose that portion of the plant under cover. Various standards of purity for potable waters have been stated and passed into disuse and in this country to-day the following guarantees by filter companies are accepted as descriptive of the purity expected.


Under conditions provided—and at the rate of filtration prescribed—“The filtered water shall be practically free from color, turbidity and matters in suspension; and when there are 3,300 or more bacteria per C. C. in the raw applied water, there shall be a removal of the same equal to an average of 97 per cent., and when there are less than 3,300 bacteria per C. C. in the raw applied water, there shall not remain more than 100 bacteria per C. C. in the filtrate.” Variations of this standard are not uncommon and at times 98 per cent, removal is demanded and obtained. The standard set by Filter Companies as to “Iron Removal” has been generally accepted and is as follows: “If the raw water contains two (2) parts per million or more, there shall be an average removal equal to 95 per cent. If the raw water contains less than two parts per million, there shall not remain in the filtrate more than an average of two parts per million. The standard for color removal is somewhat variable, but it is very acceptable if the filtrate does not show a greater amount of color than ten (10) on the platinum cobalt scale.

Platinum Cobalt Method.

Standard solution having a color of 500 is prenared by dissolving 1.246 grammes of potassiumplatinic chloride (Pt Cl4, 2KC1) and 1 gramme of crystallized cobalt chloride (CO Cl2 6H2O) in 100 C. C. concentrated hydrocloric acid, making up to 1 litre with distilled water. Through dilution. standards are prepared by mixing 1—2—and 3—C. C., etc., of this solution with distilled water in nesseler tubes to the 100 C. C. mark—about twelve dilutions usually cover the extremes ot color and read directly—1 C. C. of strong sol.= 5 p. p. m. color—2 C. C. added=10 p. p. m.—3 C. C.=15 color 4 C. C.=20. 5 C. C.=25 p. p. m. color—etc., etc. The water to be tested is compared against the standards until both the test tube of water and the particular dilute standard compare—holding both tubes over a white surface placed at an angle so that light is reflected upward through both liquids The operator looking vertically downward through the tubes. Colorimetric methods are preferred in determining iron in water and the sulpho cyanate test is usually employed. Test: 100 C. C. of water suspected is placed in porcelain evaporating dish, add 10 C. C. hydrochloric acid, then add 1 C. C. nitric acid (both acids should be iron free). Place dish over low flame and evaporate slowly to 10 C. C. (about). Allow dish and contents to cool and transfer contents to a 100 C. C. glass tube and wash dish into tube with distilled water until the 50 C. C. mark on the tube is reached Add to the mixed contents of tube, 15 C. C. of ammonia sulpho cyanate (KCNS.) and make up with distilled water to the 100 C. C. mark on the tube and compare depth of color with dilute standards made up as follows: Into each of the tubes used to contain the standards, place a known amount of “standard iron solution” making standard enough to cover the range of the particular water tested. Add a little distilled water (about 5 C. C.) add 5 C. C. hydrochloric acid, then 1 C. C. nitric acid. Dilute to 50 C. C. mark with distilled water. Add 15 C. C. ammonia sulpho cyanate sol. and make up to the 100 C. C. mark on the tube with distilled water and mix. Compare depth of color against white background. Standard iron sol, probably, had better be purchased as its production involves very accurate weighing—color comparisons should be quickly made as the color changes in the solution. The nitric acid is used to oxidize the ferrous to the ferric iron and more than 1 C. C. of nitric acid is sometimes necessary—peroxide of hydrogen is also used sometimes as an oxidizing agent. Chlorine in the liquified form or as hypochlorite of lime has become a useful adjunct to filtration and modern plants are equipped with chlorinating devices as a matter of course. Disagreeable and inefficient results follow the use of chlorine in unfiltered waters, especially if turbid or high in color, but when introduced into the filtrate in the minute quantities required, it ensures a destruction of bacteria, practically to sterilization, and gives stability to the purification process not wholly provided byfiltration unless the filter plant is very carefully operated. It is sometimes desirable to ascertain tbe presence or absence of chlorine in waters that have been chlorinated. This test is known as the “Residual chlorine test.” Great discretion is necessary in using this test as the presence of iron salts, chlorates and nitrites in water giving similar reactions may confuse the operator. The 1917 edition of Standard Methods of Water Analysis gives the Tolodin method as follows:

Reagents Used.

  1. —One gramme of 0—Tolidin purified by recrystallization from alcohol, is dissolved in one litre of 10 per cent, hydrochloric acid.
  2. —1.5 gramme copper sulphate and 1 C. C. sulphuric acid (conc.) is dissolved in enough distilled water to make 100 C. C.
  3. —0.025 grammes of potassium bi-chromate and 0.1 C. C. of sulphuric acid (cone.) is dis solved in enough distilled water to make 100 C. C.


Mix 1 C. C. of No. 1—Tolidin solution with 100 C. C. of the water to be tested in a nessler tube. Allow the mixture to stand at least five minutes— small amounts of free chlorine give a yellow and larger amounts an orange color. For quantitative work, the color obtained may be compared against standards prepared from the copper sulphate solution No. 2 and the potassium bi-chromate solution No. 3.

Chlorine in waters not subjected to chlorination treatment is indicative of animal pollution and if the treatment is indicative of animal pollution and if the samples run higher than the normal neighborhood charts—the water is under suspicion as a carrier of intestinal pollution— possibly disease producing. The regular chlorine test, therefore, offers a very valuable test as a quick way to determine the presence of such pollution on the water shed immediate to the raw water supply. In such cases, sodium chloride, the derivative of the chlorine, is not removed by filtration, being in solution, but the disease germs, the presence of which are indicated through the chlorine as established by test are removed. Then the chlorine if found is harmless and can be disregarded.

Chlorine Test: 25 C. C. of the water is placed in a white porcelain evaporating dish. Add 0.5 C. C. of chromate of potassium solution and from a graduated burette run in standard silver nitrate solution and stir with glass rod until the just faint reddish tint just appears, permanently. A comparison with similar amount of water and chromate used as a blank should be made. E. G. suppose 3.5 C. C. of standard silver nitrate solution is used then: 3.5—0.lX2=6.8—So 6.8. p.p.m. is the chlorine in the water.

Vol. 29. No. 45—page 2960 of the U. S. Public Health Report gives the U. S. standards of water to be supplied for interstate carriers as adopted October 21, 1914—and it is advisable for purveyors of water to railroads and steamships to have on file bacteriological data and conform to the standards which follow:

  1. The total number of bacteria developing on standard agar plates incubated twenty-four hours at 37 degrees Centigrade shall not exceed 100 per cubic centimeter; provided that the estimate shall be made from not less than two plates, showing such numbers and distribution of colonies as to indicate that the estimate is reliable and accurate.
  2. Not more than one out of five 10 cubic centimeter portions of any sample examined shall show the presence of organizms of the bacillus coli group when tested as follows:
  1. Five 10 cubic centimeter portions of each sample tested shall be planted, each in a fermentation tube containing not less than 30 cubic centimeters of lactose peptone broth. These shall be incubated forty-eight hours at 37 degrees centigrade and observed to note their formation.
  2. From each tube showing that more than five per cent, of the coli are in the fermentation tube, plates shall be made after forty-eight hour’s incubation upon liquid lactose litmus agar or endo media.
  3. When plate colonies resembling B. coli develop upon either of these plate media within twenty-four hours a well isolated characteristic colony shall be fished and transplated into a lactose broth fermentation tube which shall be incubated at 37 degrees centigrade for forty-eight hours.

“Examining this above recommended standard, it will be noted that roughly proportioning, twenty per cent, of the 10 cubic centimeter samples will be permitted to give positive indications of B. coli. These tests, however, are not to secure presumptive tests of the lactose peptone bile gas formation method; they are something nearer to the typical coli test. Accordingly the number of samples showing B. coli by presumptive tests may well run to forty per cent, of the 10 cubic centimeter samples. Comparing this forty per cent, standard, we note that the New York drinking water, as supplied in the 135th street gatehouse, showed a yearly average, in 1912, of fortyfour per cent, positive results in 10 cubic centimeter samples, and in 1911 an average of seventysix per cent, positive results.

“Assuming that water with forty per cent, positive results corresponds with the recommendations of the committee, the efficiency of the filtration and sterilization system, when compared with the permissible limites in raw water for filtration purposes as recommended by the International Joint Commission’s sanitary experts, ought to correspond roughly to ninety-nine per cent.”

Bacteriological tests to establish this and other standards of purity cannot be given here. They are properly within the province of the bacteriologist and chemist who is familiar with the methods used. The data itself is for the information of the producer of pure water who must comply with the legal as well as the popular requirements.

No posts to display