The recent admirable paper of Mr. Francis D. West before the American Water Works Association has covered so well the general field implied under the title of this paper that I shall confine myself to an intensive study of certain features only. The accurate determination of the ratio between amounts of liquid chlorine and hypochlorite of lime which will perform equivalent purification is not a simple matter. Previous ratios have been based upon approximate bacterial efficiency. Mr. D. D. Jackspn, in preliminary work at Ridgewood Reservoir, Brooklyn, arrived at the conclusion that the ratio was 1:9. But this work was carried out in midsummer, in hot weather, and temperature has a large effect upon the results. Under the conditions prevailing 0.13 ppm. of liquid chlorine was effective in preventing positive tests for B. coli in lOOcc. quantities of water. In experiments carried out by the Mt. Prospect Laboratory force last winter 0.35 ppm. of liquid chlorine failed to remove B. coli in lOOcc. quantity. The water, though different, was probably of sufficiently similar character for comparison. Mr. Kienle, comparing results at Wilmington. Del., gives tables supporting a ratio of 1:73.3 and states that Mr. F. H. Huy, at Niagara Falls, Western New York Company’s Plant, obtained a ratio of 1 :8. Mr. Huy later, according to West, obtained a ratio of 1:10. Mr. West of Torresdale reached a ratio of 1:6 to 1:7. Many factors must be considered in arriving at a conclusion. 1st. Overdosing will influence the results. If the dose applied is greater than required the average figure for chlorine applied will appear too large and the bacterial removal proportionately too low. Again overdosing, in my experience, frequently results in poor efficiency and should be avoided. It is my belief that it eats into sediment and sets free clumps of material containing B. coli, thus sometimes apparently increasing the B. coli and the bacteria in the treated water. From the first I have felt that the setting free of sediment was responsible for so-called aftergrowths. In fact such seems to be the consensus of opinion at the present time. 2nd. Underdosing also tends to throw error upon an average, since if the available chlorine applied is too small it produces little or no bacterial removal. 3rd. The quality of the water treated affects the results also. Turbidity and color influence the amount of needed chlorine greatly, the latter rising or falling with increase or decrease respectively. Presence or absence of microscopic organisms, hardness or softness, all have their influence. 4th. Apart from these influences the goodness or badness of the water, considered from a bacteriological standpoint, decreases or increases respectively the apparent efficiency. You all know that greater percentage removal is shown by a sand filter when the bacteria in the raw water are high, that with less than 100 bacteria in the raw water a high percentage removal is rare. 5th. A certain number of tests of treated water in my experience will show no removal or even an increase in bacteria though the dose applied is correct and sufficient. These may be due to accidental variation in the application of the chemical, the slipping by of untreated water, disturbance of sediment, etc., but they influence the average and the ratio. They occur with both liquid chlorine and hypochlorite of lime. 6th. Attenuation only of B. coli instead of destruction is frequent. 7th. The limits of analysis offer another difficulty. Ordinarily tests are made in O.lcc., lcc. and lOcc. of water. If all three dilutions are positive on the raw sample the limit may not have been reached and a test in 1cc. in the treated sample may indicate a purification of 90 per cent, when it may really be 99 per cent. Again a negative test in lOcc. in both raw and treated water may be a 90 per cent, removal although it appears to be 0 per cent, because of the lack of lOOcc. tests. 8th. Lastly the amount of chlorine actually getting into the water in hypochlorite of lime application is very irregular and uncertain, resulting in tremendous errors.

TABLE No. 1.

Comparison of Bacterial Removal at Valhalla By Liquid Chlorine and By Hypochlorite of Lime.

* Director of Laboratories, New York Department of Water Supply, Gas and Electricity. Abstract of Paper Read Before New Jersey Sanitary Association.

Results at Valhalla and Other Points.

Although the City of New York has used hypochlorite of lime and liquid chlorine at many points it has been difficult to choose a point at which the data was sufficiently exact to be reliable. For example, in the case of brooks the exact amount of water treated is not sufficiently well determined. Excess of hypochlorite of lime is fed in order to insure good bacterial results. At Smith’s Pond, Brooklyn, two sources of supply admixed, a few feet from the point of chlorination, so that mathematical combination according to volumes was necessary of the analyses of the two raw samples. At Watt’s Pond, the treated sample is mixed with well water. In the case of Croton water, treated at Dunwoodie, the treated sample taken for analysis has passed through a large reservoir so that the time element is uncertain and the daily results could hardly be studied. Only at one point did the data seem complete and sufficiently exact to compare, and even there the available chlorine is based upon the hypochlorite of lime, amount of dry powder, used. The results at this point, Valhalla, New York, with liquid chlorine and hypochlorite of lime have been carefully studied. The results with hypochlorite of lime cover February, 1912, to November, 1913, inclusive, and with liquid chlorine December, 1913, to September, 1914, inclusive. The liquid chlorine results are based on weight of chlorine delivered. The machine is an Electro Bleach Company’s with absorbing tower. Analyses have been made three times a week. The analyses for B. coli are made upon O.1cc., 1cc. and 10cc. quantities. If all three tests are positive in the raw sample and 0 in the treated the result is considered to be 100 per cent, removal. If three positive tests are reduced to one, or two positive tests to none, the removal is considered to be 99 per cent. If three positive tests are reduced to two, two to one, or one to none, the removal is considered to be 90 per cent. In averaging several B. coli results to find percentage reduction 0.1cc. results are multiplied by 100, lcc. results by 10, and added to the lOcc. results. The resulting sums are then compared. The average analyses at Valhalla are given in Table 1:

In order to study the results more closely, the analyses were grouped by amounts of chlorine applied and separately averaged. The results obtained by liquid chlorine between 0.14 to 0.36 ppm. appear to be justly comparable to those obtained with hypochlorite of lime between 0.41 to 0.63 ppm. Tabulated these averages are as follows m Table 4:


Comparison of Bacterial Removal at Valhalla By Liquid Chlorine 0.14 ppm. to 0.36 ppm. and by Hypochlorite of Lime, Available Chlorine 0.41 ppm. to 0.63 ppm.

(—% removal—s

The ratio of liquid chlorine to available chlorine of hypochlorite of lime is 48 per cent. In order, if possible, to make a still more exact comparison, the averages were obtained of the ppm. of chlorine which produced respectively 100 per cent., 99 per cent., 90 per cent., 0 per cent, and (—) per cent, results for B. coli. These averages are given in Table 5:

TABLE No. 5.

Ratio of Liquid Chlorine to Available Chlorine of Hypochlorite of Lime at Varying Percentages Removal, Valhalla.

* 68% of these samples could show only 90% removal, as there was not sufficient B. coli in the raw water.

The average amounts of chlorine in the above table are about the same for different degrees of purification, again indicating no beneficial effect from overdosing of chlorine, since the higher amounts do not show better percentage removal. That a large amount of the available chlorine of hypochlorite of lime may be wasted has been demonstrated this year at our Dunwoodie plant. West states that, owing to variation in strength (21 per cent—43 per cent, of available chlorine in the powder), lumpy condition, difficulties of admixture, and variations in feed, under best conditions only 87 per cent, of added available chlorine is obtained at Torresdale. At Dunwoodie there are two 5,000-gallon concrete tanks into which alternately a drum of hypochlorite of lime (700-800 pounds) is washed by water delivered with force from hose. The emulsion passes through a coarse screen where lumps arc crushed with a hoe by hand, and the solution is then agitated with compressed air. Improper bacterial results were obtained in June which led to a thorough investigation by the Laboratory force. Analyses were made every one-half hour of the hypochlorite of lime solution and it was found to be very irregular in strength and to contain not nearly what it should of the hypochlorite of lime added. As low as 10 per cent, of the available chlorine added was delivered at times toward the end of a run, and the average loss was altogether too high. Instructions were given the laborers to crush more thoroughly all lumps, to blow the solution several times during the rest period and finally settle two hours. The air blast is applied three to five minutes at a time. Later the system was adopted of blowing the sludge in a half tank of fresh water before adding fresh hypochlorite of lime. When heavier dosing is necessary, because of a large amount of B. coli in the raw water, two drums are successfully added to each tank, making over 1 per cent, solution of available chlorine. This seems preferable in order to give sufficient time for solution and settling, a tank being emptied in six hours instead of three hours when feeding eight drums a day. The sludge is washed out after an accumulation of about 30 drums when there is an unavoidable loss of some of the solution. Samples at the beginning, middle and end of each run have been analyzed since the middle of July and serve as a good check on the laborers who make up the solution. The first week about 95 per cent, of the chlorine added was actually applied, the second week it dropped to 85 per cent., and the third week to 75 per cent. Whenever a poor run is called to the attention of the laborers results improve. Chart No. 4 shows several days’ results. When the three analyses for a single run vary there has been insufficient blowing. The results at midnight of July 17 were due to a heavy thunder storm during which, the men afterward admitted, they had failed to use the electric pump which drives the air. Absolute changes of the value between tanks is due to carlessness in breaking up lumps, variation in solubility of the hypochlorite of lime, differences in weight of the powder in different drums, and variation in percentage of available chlorine. To equalize these fluctuations the water of the new aqueduct (seven-eighths of the supply) is passed through Jerome Park Reservoir entering at the lower end and leaving at the upper middle side so as to create circulation, admixture and storage. The storage capacity of the reservoir is about three days’ supply. The most interesting feature of this work, apart from economy, is the reduced amounts of hypochlorite of lime used to produce such good results. Some of the individual figures were as low as 0.10 to 0.14 ppm. chlorine and at present daily averages are between 0.25 and 0.40 ppm. This is about the quantity at which we have maintained our liquid chlorine figures in treating Bronx Water at Valhalla, of quite similar character. There have been published figures for available chlorine of hypochlorite of lime treatment effective at 0.1 ppm. chlorine. If based upon sufficiently accurate data, they tend to support the above remarks and show that hypochlorite. of lime may be active in as small amount as liauid chlorine Cincinnati, Ohio, has used liquid chlorine on filtered water as low as O.lSppm. Pittsburgh, Pa., has utlilized hypochlorite of lime, introduced directly into the water so that it was completely available, on filtered water as low as 0.12ppm. available chlorine. During the spring of the present year an extended series of experiments were made comparing the application of liquid chlorine to water by direct application and by previous solution in running water. Two machines were used, the Leavitt-Jackson with corundum distributors and the Electro-Bleach with an absorbing tower. The experiments were made at Smith’s Pond, Brooklyn, on a mixture of water from Smith’s Pond with conduit water, the latter a mixture of surface and well water. The two raw waters were analyzed separately and results combined proportionately to the amounts of each. Each machine was run 24 hours alternately and three samples for analysis, one-half hour apart taken during the last hour and second hour of each run, so as to bring the character of the water compared as near alike as possible. The Table 10b gives a summary of the results:


General Averages Smith’s Pond Test.

It appears from these results that the chlorine yields the same results bacteriologically whether introduced directly to the water or put into solution first. Bacteria counts were made on both gelatin and agar. An interesting feature is indicated in the fact that the absolute number ot bacteria growing upon gelatin at 20°C. and upon agar at 37°C. were practically the same in the treated water, tending to indicate that the residual bacteria were of a type that grew either at 20 or 37°C. and were probably identical. Tests were made for B. coli in the lOOcc. quantity also. With one exception this test was always positive in the treated water, possibly due to the fact that cold weaher prevailed. In order to study the rapidity of the action of the chlorine a series of square holes were made into the aqueduct, lined with boards, and fitted with a cover in order to avoid contamination in sampling. There were six holes, the last several hundred feet from the machines. Samples were taken, tested with orthotolidin and compared with standards. About 0.2 ppm. chlorine was immediately consumed. Hole No. 2 was only a few feet away from the point of treatment. Such immediate consumption of chlorine probably varies with the character of the water treated. The water in these tests was moving at the rate of about three feet per second. That the action of the chlorine is practically instantaneous is also shown by the fact that the bacterial results at an intermediate point of sampling within a few feet of the point of treatment averaged practically the same as at the furthest point of sampling as follows:

Both machines showed fluctuations in amount of chlorine fed as related to that indicated. The Leavitt-Jackson varied about + 5 per cent, but fed a total amount practically as indicated. This machine delivered a definite weight of chlorine, consequently pressure and temperature has no effect except as regards lag in adjustment automatically made by the machine when conditions are fluctuating. The Electro-Bleach yielded less weight of chlorine than indicated, sometimes as low as 72 per cent. This is in keeping with results at Somersworth, N. H., ‘a. In a run from March 31 to June 26, feeding 0.34 ppm. chlorine, the actual feed was 73 per cent, of the indicated amount. In another run from June 26 to August 28, feeding 0.38 ppm., the actual feed was 83 per cent, of that indicated. In testing the machine at Smiths’ Pond, the water flowing through the tower was measured by means of a 1/8-inch orifice and a mercury “U” tube. The solution was analyzed for available chlorine.


1. Intensive study of the application of liquid chlorine and of hypochlorite of lime during the course of three years at Valhalla, N. Y., points to a ratio of 1:7, chlorine to hypochlorite of lime, as productive of equal bacterial purification when hypochlorite of lime is applied under usual conditions. 2. Insufficient attention to details, causing serious loss of available chlorine, is a large factor in the use of hypochlorite of lime, and greatly affects the above ratio. When efficiently used the ratio approaches 1:3. 3. Liquid chlorine is equally efficient whether applied to the water as a gas or in solution. 4. A certain immediate consumption of chlorine takes place, probably dependant on the character of the water treated. This is true with both liquid chlorine and hypochlorite of lime. Liquid chlorine soon disappears entirely from the water whereas residual hypochlorite of lime may persist for weeks. 5. The bactericidal action of chlorine is practically instantaneous. 6. The amount of chlorine treatment should always be based on analysis of the solution with hypochlorite of lime and upon loss in weight of chlorine with liquid chlorine. I may perhaps in closing best add in concise form the important advtantages liquid chlorine has over hypochlorite of lime: 1. Compact installation. 2. Ninety per cent, less storage space required for chemical. 3. No loss of strength of chemical during storage. 4. No sludge problem. 5. No residual hypochlorite in water, hence less corrosive effect, no complaint of odor, and less danger from overdosing. 6. More effective per unit applied. 7. More uniform application. 8. Better and easier regulation. 9. Easier to handle generally and less obnoxious. 10. No increase of hardness in water. 11. Labor cost less and chemical more efficient making total cost about equal.

No posts to display