1896.] MICROSCOPICAL JOURNAL. 91 



washed by a reverse current which caused the quartz to 

 boil. The agitation and friction of the particles were in- 

 creased by means of a rake with long teeth which re- 

 volved about a central column in the filter ; the rake 

 penetrating the bed by a screw motion from top to bottom. 



From the various kinds of coagulant or precipitant 

 used, basic sulphate of alumina was selected as being the 

 most satisfactory and effective and was used in all the ex- 

 periments mentioned. The amount of alumina used was 

 J grain to the gallon of water filtered, a lesser quantity 

 failing to satisfactorily remove the organisms. The 

 amount of f or one grain per gallon did not increase the 

 removal of the bacteria, while the efficiency of the filter 

 was greatly decreased by reducing the amount of the 

 floAv through the filtered bed. 



The alumina was applied in a free flow at the beginning 

 of a run by pouring into the filter, as the water entered, 

 a pint of the coagulant containing about 911 grains of 

 sulphate of alumina for an average fiow of 1-28,000,000 

 gallons per acre. The solution was made by adding one 

 part of the alumina to six parts of water; as a result of 

 this addition there forms a white fiocculent precipitate 

 over the surface of the grains of quartz and is the actual 

 medium through which the filtration takes place, the 

 quartz serving merely as a supporting bed or sieve. 

 It required about six minutes to form this layer. When 

 applied at the rate of a drop at a time and not in a "free 

 flow" it required about a half an hour before the filtering 

 layer would be formed. As soon as the filtering layer 

 was formed the alum solution was dropped in continously 

 during the run from a regular stop at the rate of a drop 

 a second. The efl'ect of the presence of this layer was 

 to reduce the head or pressure .28 of a foot for 128,000,- 

 000 gallons per acre. The depth of the water above the 

 bed af the commencement of the run was nine inches; the 

 average length of the rurj. was about eighteen hours. 



