determination of species have shown, however, that the numbers observed in the effluents are greater than the numbers actually passing the filters. It was found, for example, in Filter No. 8, in November and December, 1891, that the most abundant form in the applied water was totally absent from the effluent; and that the species found in the effluent were very vigorous forms, capable of growing on an extremely small food supply, and that they probably grew in the underdrains and did not come from the applied water. From time to time pure cultures of the typhoid bacillus and of other more easily recognizable bacteria have been applied to the filters, and results as to their passage obtained. In the following table is given a summary of the results of such experiments as far as the data are complete : Table showing the Removal of Bacteria by Different Filters. In some of the earlier experiments the quantity of food material applied with the bacterial dose was so great as to support a vigorous growth of bacteria in the lower part of the filter and in the underdrains, greatly increasing the numbers observed in the effluents in those cases when the removal was not complete. In many cases the doses applied contained a vastly greater number of bacteria than could ever exist in a practical case, sometimes exceeding a million per cubic centimeter in the applied water. With these excessive doses a much larger percentage of the bacteria are able to pass; and from the later experiments, where these conditions were avoided, we may safely conclude that in actual practice, with smaller numbers of bacteria constantly added with the applied water, nothing like the percentages given in a few of the above experiments would be able to pass the filter. The general result of these experiments, as far as completed, is to show that, with the same material and rate, the deeper filters are much safer than the more shallow ones; that low rates are safer than high rates, and that, at the same average rates, continuous is quite as effective as intermittent filtration in the removal of bacteria. The formation of a surface film or covering, regarded by Piefke as of prime importance, appears, thus far, to have had very little influence on the results. Further discussion of these and other points is postponed until more complete data are obtained. The results already obtained clearly show that, with due care in the selection of the filtering material, filters can be constructed and operated at rates up to one and one-half to two million gallons per acre daily, with the removal of substantially all of the bacteria. These results, taken in connection with the remarkably low death rates from diseases known to be capable of conveyance by drinking water in some European cities having filtered water supplies, from polluted sources, leave no room for doubt as to the efficiency of suitable filtration as a safeguard against water-carried diseases. * Zeitschrift für Hygiene, 1889, vol. vii., p. 115. FILTER TANKS Nos. 8, 18 A AND 20 A. The earlier histories of these filters were given in the special report upon the Purification of Sewage and Water, pages 602, 658 and 695. Filter No. 8 is filled with No. 6 sand, with a layer of loam eight inches thick, the top of which is six inches below the surface; No. 18 A is filled with No. 1 sand; and No. 20 A with No. 1 sand, with a layer of No. 2 sand one foot thick, the top of which is three feet below the surface. The mechanical analyses of the sands were given on page 429. Filter No. 8 has continued to give from the Merrimack River water (city water until Nov. 12, 1891, and afterward canal water) an effluent which compares favorably with the best spring waters, being practically colorless and free from objectionable organic matters and bacteria. Filter No. 18 A in 1890 received city water at an average rate of 819,000 gallons per acre daily, applied in hourly doses for fourteen hours in the day, and six days in the week. In October the effluent commenced to be colored, and during November no water was applied. In December water was again applied as before, but the color of the effluent was not lower than before the rest. During 1891 water was applied in smaller doses, each equivalent to 5,000 gallons per acre, aggregating as nearly as possible a million gallons per acre daily. The single doses followed each other with intervals. of only about seven minutes, days, nights and Sundays. Commencing Aug. 18, 1891, canal water was applied instead of city water. Filter No. 20 A received city water during the entire time. In the latter part of 1890 the fine No. 2 sand three feet below the surface became clogged, so that water stood in the coarse sand, making the filtration practically continuous. On December 23 a small reservoir with a siphon was provided, which delivered a dose of water equivalent to only 10,000 gallons per acre, and doses were applied as frequently as possible, without causing water to remain standing in the coarse sand. The average results obtained from these filters, by periods, from the times when water was first applied to them, are shown in the following tables: |