tion the changes of the water and the personal equation are not such important factors, as the sand filter, operating at a low rate, adapts itself more readily to changes in the character of the raw water, receives no coagulant, and its efficiency rests on natural rather than chemical means. It is, of course, true that on this account a slow sand filter is more easily worked by unskilled hands and better results obtained than when a mechanical filter is operated by those insufficiently trained to appreciate the necessity of constant care and attention. This gives a certain percentage of safety to the slow sand filter. A decrease of filter efficiency, even for a few minutes, may, of course, be of serious moment when treating a polluted water, and especially if these few minutes occur several times in the course of a day or of a week. It is the aim at the present time of those planning mechanical filters of the modified type, used for the purification of public supplies, to so construct them that failures are almost impossible. Constant examination of the raw water and effluent is also insisted on. It is undoubtedly true that, with expert supervision, a mechanical filter of the latest design is capable of giving a bacterial efficiency nearly equal to, but probably not quite as great, as that obtained by sand filters. Whether this very slight difference in efficiency is of any hygienic importance or not yet remains to be demonstrated.

I have collected some data bearing on this phase of the water filtration problem, the results covering thirteen different experimental filters and two large filtration systems. In every case the period that I have selected for comparison of these filters has been such that each, as far as I can determine, was being operated under normal conditions, and all analytical results obtained during this period are included in this comparison. From the figures obtained it will readily be seen that the efficiency of all these filters was at times lower than the normal and desired efficiency. Taking the results of the two types of filters collectively, the slow sand filters had an efficiency of over 99 per cent. 42 per cent. of the time, of over 98 per cent. 64 per cent. of the time, and below 95 per cent. 13 per cent. of the time; while the mechanical filters had an efficiency of over 99 per cent. 32 per cent of the time, of over 98 per cent. 49 per cent. of the time, and less than 95 per cent. 28 per cent. of the time. The filters used in this comparison are the Lawrence city filter, the Little Falls filter, and experimental mechanical and sand filters at Lawrence, Washington, Cincinnati, Pittsburg and New Orleans.

Finally, the chief test of the efficiency and value of filtration is the effect that filtered water has on the health of a community using such water, compared with the health of the same community when using polluted

water before the introduction of filters, or, of course, it can be shown also by comparison between communities using purified water and those still using polluted water. Unfortunately, few results are obtainable showing the direct effect on the health of a city of mechanically or chemically filtered water, while we have, on the other hand, many illustrations, especially abroad, of the hygienic efficiency of slow sand filtration, and in this country the results of years of operation of the Lawrence filter and Albany filter. The Lawrence results have been related so frequently that many of you are familiar with them, but, as they are as yet the foremost American exemplification of the value of sand-filtered water, they must be repeated here.

The water works of the city of Lawrence were built in 1875, water being taken from the Merrimack River, and soon after their introduction the number of deaths from typhoid fever in the city began to increase. It was finally recognized in the last of the '80s that whenever there was an epidemic of typhoid fever in Lowell, which occurred practically every year, there was also an epidemic of typhoid fever in Lawrence, the city of Lowell, with its present population of 100,000 people, being located about nine miles up the river from the intake of the Lawrence water works, and all its sewage passing into the Merrimack. Investigations showed that infection from Lowell reached Lawrence and was distributed throughout the Lawrence water supply system, and this investigation resulted in the construction of the Lawrence filter in 1892 and 1893.

In the tables given, showing the growth of the city of Lawrence in population, the total number of deaths in the city, year by year, is also shown, and the total number of deaths from typhoid fever before and after the construction of this filter. It will be seen from this table (Table 1) that the greatest number of deaths from typhoid fever in the city, namely, 60, occurred in 1890, when the population of the city was less than 45,000; that in 1892, the year before the construction of the filter, the total number of typhoid deaths was 50, the city having in that year a population of about 48,000; that in 1894, the first year that filtered water was supplied to the city, the number of deaths from typhoid decreased to 24 (more than in any subsequent year to date, as shown by the table), and the total deaths in the city decreased more than 300, or from 1,211 to 901. In 1906 the total number of deaths was 1,330, an increase of 9.8 per cent., and the population of the city was 76,000, an increase of about 75 per cent. The total number of deaths from typhoid fever in 1906 was 15, as against 50 in 1892, making the comparative death rate from typhoid fever as follows: in 1892, 111 per 100,000, and in 1906, 20 per 100,000, or a decrease of 83 per cent. The rates for the intervening years are shown on the table. In Albany

the reduction was from an average of 71.5 deaths per year from typhoid for the nine years preceding the building of the filter to an average of 23.8 for the six years following the construction. The reduction in the number of cases of typhoid in Albany was from an average of 424.8 cases per year for the nine years preceding the building of the filter to an average of 100.5 for the six years following construction. The total deaths also decreased very remarkably as at Lawrence.

TABLE 1.— Vital Statistics, Lawrence, Mass., 1888 to 1906, inclusive.

TABLE 2.- Number of Deaths from Typhoid Fever reported in Albany during the Nine Years before and Six Years since the Starting of the Filter Plant.1

While, as I have stated, the actual hygienic efficiency of modern mechanical filters has not been as yet adequately demonstrated, yet such efficiency is, I believe, undoubtedly obtained and will be demonstrated

1 Compiled from records of Bureau of Health and State Board of Health.

eventually. There is a place for each type of filter. Sand filters are successful with polluted but comparatively clear waters, and, with the addition of sedimentation basins and the occasional use of coagulants, with waters somewhat turbid. At the present time, however, no method of satisfactorily handling the very turbid waters full of clay and other matters in suspension is known other than by the use of coagulants and mechanical filters, although sedimentation and double sand filtration are of much promise. The prejudice in some minds against the use of aluminum sulphate under proper supervision is, I believe, not justified; the aluminum hydrate formed is invariably removed by the filters if they are successfully operated. We might, perhaps, as well refuse to eat many common foods because in their production chemicals are used as to object to water clarified by coagulants. The public is beginning to demand a clean water of good appearance, as well as safe, and in order to obtain such a water coagulants are sometimes necessary. Before either type of filter is adopted, thorough preliminary studies of all the local conditions as to water supply, etc., should be made, and hasty and inexpert conclusions prevented.

TABLE 3.- Number of Deaths from Typhoid Fever and Rate per 100,000 Population in Albany and Neighboring Cities.1