CHAS. K. FOWLER. 291 



A marked variation is observed in the number of 

 bacteria in the river water, doubtless owing, in part, to 

 different stages of the tide. The bacteria of the last 

 samples are noted as "liquefying," so that they probably 

 do not represent the full numbers present. 



When the beds are in good order the efficiency is very 

 good, considering the rate of filtration. The difficulty, 

 under present conditions, is to keep them in good order 

 and obtain a sufficient supply of water. 



This has been impossible during this winter, or, in fact, 

 in any winter, though in some a much nearer approach 

 can be made than in others. 



The longer the beds can run without cleaning or 

 undue pressure, the more efficient they are. Results 

 immediately after cleaning are not as good as a week 

 later. 



Coagulants can, of course, be used with this system of 

 filtration; but alum has never been used because of the 

 belief that, in the quantity required by the water of the 

 Hudson, its use would be more prejudicial to health than 

 the turbidity. An attempt was made by a private com- 

 pany, about three years since, to introduce oxide of iron 

 by the method before described, but thus far it has 

 failed, by reason of mechanical difficulties. 



It is said to have been very successful in connection 

 with some of the European filter plants. Here, however, 

 the experiments made indicate that long contact with 

 the iron is required, with large sedimentation capacity 

 before passing to the beds, and filtration at a very slow 

 rate, to effect satisfactory results. In fact, all experience 

 thus far points to the conclusion that the Hudson River 

 water is not adapted to the use of coagulants. 



I am convinced that the best method of purifying the 

 waters of the Hudson at Poughkeepsie is that of slow 

 sand filtration, as now in use. One essential, however^ 

 is plenty of filtering area, so that a sufficient surface may 



22Q 



