WATER FILTRATION. XXIII. 



have an apparatus enabling one to take specimens of tlie 

 water immediately after its leaving the filter. 



3. Filtered water containing more than 100 germs capable 

 of development per cubic centimetre, must not be allowed 

 to enter the pure water reservoir. The filter must there- 

 fore be so constructed that insufficiently cleansed water 

 may be removed without mixing with the well filtered 

 water. 



Dr. Koch further states: — "To these sentences I have 

 some further remarks to add. Strictly speaking the last 

 two demands would alone suffice to avert the danger of 

 the infecting of filtered water so far as is possible with sand 

 filtration at all. But it seems to me questionable whether 

 the demand of the daily bacteriological control could be 

 limited to times of danger, that is to times of maximum 

 consumption of water, periods of frost, and when epidemics 

 seem imminent; in the intervals a less elaborate control, 

 say the examination of collected waters every three days 

 would suffice. The weekly bacteriological investigations 

 of the collected filtered water, now usual where the water 

 is bacteriologically examined at all, is to be regarded as 

 insufficient under all circumstances. For the times of less 

 bacteriological control, however, guarantee for the regular 

 working of the filters must be given by strict limitation and 

 control of the rate of filtration. The rate of filtration is 

 generally calculated by ascertaining the proportion of the 

 water filtered in 24 hours to the filter. But everybody who 

 knows the ordinary working of water works knows also the 

 demands made upon them in the course of 24 hours vary 

 very considerably. At certain hours of the day very much 

 water is consumed ; at night, on the other hand, the con- 

 sumption is small. If the reservoir is not large enough to 

 balance these inequalities, this is effected by changes in 

 the rate of filtration. The statement then that a water 



