EXAMINATION OF FILTERS 313 



containing not more than 100 bacteria per c.c. These 

 coefficients, however, take no account of the class of sand 

 used or character of water filtered, and they are no longer 

 regarded as trustworthy. When a filter-bed is freshly con- 

 structed, organisms are washed through it with great 

 rapidity, but after a certain quantity of water has passed 

 through, or the water has been allowed to stand upon it for 

 a certain time, a slimy coating of detritus and bacteria is 

 formed on the surface. If water is slowly passed through 

 the filter when this coating has been formed to a sufficient 

 extent, which will occur after a period varying mainly with 

 the composition of the water, the majority of the bacteria 

 will be retained by this surface, either by sticking to it or 

 by being strained off. The increasing thickness of this 

 coating wUl reduce the velocity with which the water passes, 

 and at the same time some of the bacteria will tend to grow 

 downwards into the lower strata of the filter, and, if the 

 process were continued long enough, would be washed 

 through into the filtrate, and ultimately become more 

 numerous there than in the unfiltered water. The in- 

 creasing resistance to the passage of water would also make 

 it necessary for the pressure to be increased, which would 

 in this class of filter assist the passage of organisms. It is 

 therefore necessary in the working of sand-filters to run the 

 filtrate of each bed to waste, or to permit a body of water to 

 stand on the filter without filtration, until a sufficient 

 coating has been formed to arrest organisms ; to stop 

 filtration when the deposit has increased to such extent as 

 to threaten the renewed passage of organisms ; and to 

 remove the upper or filtering layer, and permit a fresh 

 deposit to be formed. The indication for scraping usually 

 adopted is that the filter-bed no longer passes the required 

 quantity of water under the maximum permissible head. 

 The sufficiency of this practice has not been clearly shown. 



