90 THE AMERICAN MONTHLY [Marcli 



The mec'hanical form was represented by one of the New 

 York Filter Company's filters and one of the so-called 

 Morrison filters. After the first seven months the sand 

 filters were discontinued, it having been satisfactorily 

 ascertained that the length of run was much less than the 

 mechanical filter before the bed became clogged and the 

 rate of flow in the natural bed was but 30,000,000 gallons 

 per acre in twenty-four hours, while the mechanical filter 

 was run at the rate of 125,000,000 gallons per acre in 

 twenty-four hours. The efficiency of removal of bacteria 

 was not as high, and the results were variable, either as 

 the result of cracks in the filter or from some unknown 

 cause. Although both of these natural filtration beds were 

 constructed exactly alike, the results from the second were 

 much poorer than from the first. When the natural bed was 

 transformed or assisted by the addition of alum, thus 

 converting it into a mechanical filter, the removal of bac- 

 teria was increased to nearly the same as on the Morrison 

 filter, but the length of the run was correspondingly de- 

 creased. 



The sand used in the natural beds Avas a natural river 

 sand, not over sharp, while the sand used in the mechan- 

 ical filter was crushed quartz having sharp edges. 



In the beginning of the experiments, the New York filter 

 gave such varied and unreliable results that its use was 

 abandoned, while the so-called Morrison filter was con- 

 tinued in use during the whole series of experiments, 

 which lasted for a period of about ten months, the work- 

 ing- of the mechanical parts of the filter being perfectly 

 satisfactory and the results obtained being successful. 



The filter bed used in this mechanical filter was two 

 feet and ten inches in depth, supported upon a base of iron 

 with circular perforations of about 4 inches in size, which 

 were covered with screens. The crushed quartz used was 

 the "effective size" of 0.59 millimeters. The filter was 



