As the carbonate falls to the bottom of the tank it carries down with it the organic particles. Hence sedimentation is brought about by means of chemical precipitation. It is obviously a mechanical process as regards its action upon bacteria. Nevertheless its action is well-nigh perfect, and 300 or 400 m.-o. per cc. are reduced to 4 or 5 per cc. We shall refer to this same action when we come to speak of bacterial purification of sewage. _Alum_ has been frequently used to purify waters which contain much suspended matter. Five or six grains of alum are added to each gallon of water, with some calcium carbonate by preference. Precipitation occurs, and with it sedimentation of the bacteria, as before. But, as Babes has pointed out, alum itself acts inimically on germs; in such treatment, therefore, we get sedimentation and germicidal action combined. As a matter of actual practice, however, sedimentation alone is rarely sufficient to purify water. It is true that the collection of water in large reservoirs permits subsidence of suspended matters, and affords time for the action of light and the competitive suicidal behaviour of the common water bacteria. Yet, after all, filtration is the most important and most reliable method. _Sand Filtration_, as a means of purifying water, has been practised since the early part of the present century. But it was not till 1885 that Percy Frankland first demonstrated the great difference in bacterial content between a water unfiltered and a water which had passed through a sand filter. Previous to this time the criterion of efficiency in water purification had been a chemical one only, and the presence or absence of bacteria in any appreciable quantity was described, not in mathematical terms, but in indefinite descriptive words, like "turbid," "cloudy," etc. It is needless to say that this difference in estimation was due to the introduction by Koch of the gelatine-plate method of examination. As a result of Percy Frankland's work, he formulated the following conclusions as regards the chief factors influencing the number of microbes passing through the filter. It depends upon: (1) _The Storage Capacity for Unfiltered Water._ This, of course, has reference to the advantages, which we have noticed above, of securing a large collection of water previous to filtration for subsidence, etc. (2) _The Thickness of Fine Sand through which Filtration is Carried on._ An argument needing no further support, for it is clear, other things being equal, the more sand water passes through the greater the opportunity of leaving its impurities behind. (3) _Rate of Filtration._ The slower filtration will be generally the more complete in its results.