ACTION AND UTILITY OF FILTERS. 281 



important to feed the vessel containing the emulsion with water 

 so as to prevent the bougie from becoming dry. The filtrate 

 which collects in the exhausted flask is then examined in the 

 usual manner. In the case of filters which require considerable 

 pressure, the receptacle for the unfiltered water must be filled 

 with the special emulsion of a non -pathogenic organism, and 

 then connected to a service supply or to a tank raised sufficiently 

 to give the required head of pressure. When the delivery of 

 filtered water is very rapid, it is necessary to collect a considerable 

 amount of the filtrate, especially during the first few minutes of 

 working, as Gruber has shown that this is the critical time of the 

 experiment. I have found that under these conditions it is best 

 to pump several litres of the filtrate through a Pasteur-Cham ber- 

 land candle and diffuse the deposit in 5 c.c. of sterile water. The 

 whole of this mixture should then be plated out in gelatine, 

 gradually increasing amounts being used for each plate. 



In order to test the indirect passage of micro-organisms 

 through the filter the plan already described under the experi- 

 ments made with the Berkefeld candles is very convenient. If, 

 however, there is no mantle supplied with the filtering candle 

 this must be connected to a partially exhausted sterile Kitasato 

 flask by means of stout rubber fitted with a pinch-cock. The 

 candle is then placed in a narrow-necked jar, containing either 

 an emulsion of B. typhosus or Spirillum Cholera? Asiaticse, in 

 sewage or polluted water. The filtering portion of the candle 

 only must be placed in the fluid, and the neck of the jar is then 

 plugged with sterile wool. By releasing the pinch-cock a few 

 c.c. are allowed to filter through the candle every day, taking 

 care to replace the amount of fluid filtered so that the surface 

 of the candle may not become dry. The filtrate in the flask 

 in the case of the B. typhosus should be mixed with a large 

 quantity of broth, and then incubated at 37 C. ; if a turbidity 

 occurs, loopfuls should be plated out and tested in the usual 

 manner. In the case of the Spirillum of cholera the filtrate 

 should be converted into 1 per cent, peptone and | per cent, salt 

 solution by adding a sufficient quantity of the stock peptone 

 and salt solution. The flask is then incubated at 37 C., and 

 loopfuls removed from the surface after ten, eighteen, and 

 twenty-four hours are examined in the usual manner. 



