The effect of this filtration is chiefly _mechanical_; it separates any suspended matter, whether organic or inorganic. A certain amount of _biological_ action possibly also takes place. Piefke found that a perfectly cleaned and sterilised filter when first used, increases the microbes in water, instead of decreasing them. Gradually a gelatinous layer of slimy matter is formed on the top of the sand; the water now filters through much more slowly, but it gradually becomes freer from microbes, these being intercepted by the slimy layer. It is important that this layer should not be disturbed by too rapid or forced filtration, and that when the surface layer requires to be removed, because the filter has become impervious, time should be allowed for another thin film to form before the filtered water is again utilised. Koch concluded that the rapidity of filtration should never be allowed to exceed 100 millimetres (about 4 inches) per hour; and that the number of microbes per c.c. in the filtered water should never exceed 100. Some oxidation of organic matter, as well as detention of microbes, may take place during the filtration of water, nitrates being formed by the vital activity of certain “nitrifying” microbes in the filter. (On nitrification, see pages 195 and 274.) P. Frankland’s observations show that the number of microbes in Thames water is reduced by filtration through sand and gravel beds, as practised by the London Water Companies, so that only 3·4 per cent. of those originally present remained. He also concludes that the majority of the microbes present in filtered water are derived from post-filtration sources. Thus the number is greater in tap-water than in water derived from near the reservoirs. Other materials besides sand have been used for filtration on a large scale, but none with proved success. =Domestic Filtration= ought, as already explained, not to be needed, but circumstances often arise in which the public supply is open to suspicion, and a second domestic line of defence against infection through the water supply is desirable. When this is so, the form of filter which will best protect the household is one attached to the house-tap, so that all drinking-water is perforce filtered. When filtering involves the transfer of water from the tap to the interior of the filter, opportunity is left for carelessness or forgetfulness. The one essential point of a domestic filter is that it will prevent the passage through it of microbes. Every filter must be tested from this standpoint. On this point the experiments of Woodhead and Cartwright Wood are conclusive. They first of all experimented on various filters with fine artificial ultramarine containing particles 16 µ to 0·6 µ or even less in diameter in suspension; and milk containing granules and globules of fat 0·5 µ to 30 µ or more in diameter, freely diluted with water.