PURIFICATION OF WATER FOR DRINKING PURPOSES 165 



the great difference in the numbers of bacteria found in 

 the effluent from day to day (see Table). 



Experiments were made to determine whether the 

 increased number of bacteria with the greater flow 

 was due to an increased number coming through the 

 sand of the filter at the time, or was due to washing 

 out, in consequence of the greater flow, numbers that 

 had accumulated or grown in the under-drains or gravel 

 on the bottom exposed to air entering the outlet pipe. 

 For this purpose the outlet pipe was closed, when water 

 was slowly coming through the sand, for several hours, 

 and the accumulated water drawn out at varying velo- 

 cities. It was found that when this accumulated water 

 was drawn at a slow rate very few bacteria were in it, 

 and that the number increased greatly when drawn at 

 & rapid rate, showing that some bacteria were then 

 washed off from their resting-places in the drains and 

 in the gravel ; but this increase in the number of bac- 

 teria was not as great as that which was observed when 

 the effluent was coming directly, at the greater rate, 

 through the sand. Numerous other experiments were 

 made, which proved conclusively that the great increase 

 in the number with the increase in rate was due to 

 more bacteria actually passing through the sand with 

 the increased rate of flow. 



The sand of this tank was examined for bacteria by 

 boring a hole from top to bottom and taking out sam- 

 ples at different depths and determining the number of 

 bacteria in 1 gramme of the sand derived from different 

 depths. It is possible that in obtaining the specimen 

 from the lower part of the tank some uncertainty may 

 exist in the results from sand falling from the upper 

 layers. The following table gives the results of these 

 observations : 



