At Chesterfield, contact beds and an intermittent percolating filter worked by means of a Shone's ejector were put down side by side. They were both constructed of destructor breeze, which was crushed to the same size, namely,

inch to about inch. The albuminoid ammonia in the effluent from the percolating filter was about 0.06 to 0.07 parts per 100,000, while that in the effluent from the contact beds was over 0:11. The Sewage Committee were so satisfied of the superiority of percolating filters that they have ceased to use the contact beds.

Conclusions with regard to Contact Beds.-Where sewage is previously submitted to liquefaction in a septic tank, either open or closed, or a subsidence tank, in conjunction with a lateral or upward-flow bacteria bed, a satisfactory effluent can be obtained by means of double contact beds worked with three fillings a day. If the plant for the preliminary treatment of the sewage is curtailed, contact beds can only be relied upon when worked with two fillings a day, and consequently the area to be provided must be increased.

When, however, the sewage for any particular reason is exceptionally weak, and an adequate septic tank space is provided in conjunction with a small upward-flow or lateral filter, single contact beds may be sufficient, if the sewage is properly distributed and the filter material is small.

Percolating Filters.-The Commissioners in their Report speak of two artificial filtration processes-contact beds and continuous filtration.

In this volume I have adopted the phrase "percolating filters," instead of that of "continuous filters," because some of the continuous filters are worked intermittently, and intermittent continuous filtration is a verbal contradiction. Doubtless the expression "continuous filtration" has been adopted in contra - distinction to contact beds, the outlets of which are closed, and which have two or three fillings a day. With continuous or percolating filtration the outlet of the filter is open all the time, the sewage being distributed over the surface of the filter bed through perforated pipes or some other arrangement; it percolates uninterruptedly through the filter and out at the bottom, the process taking place in the free presence of air. To enable as much air as possible to

enter the filter, the supply of the sewage to the filter is frequently made intermittent, the sewage being distributed over the surface of the filter in thin films or single drops. The films of sewage soaking down into the filter displace the air in front and draw air in after them. This intermittency of action is to be encouraged in every way. It appears to me that the expression "intermittent percolating filters" most correctly describes the arrangement, which, I believe, gives the best results.

The process of purification by intermittent filtration can best be understood by a careful study of the Chesterfield experimental filters, shown in Fig. 18. The filters here are two in number, each with an area of 616 square yards, the area of the two being a little over a quarter of an acre. Between the two filters is a Shone's ejector, from which an iron main supply pipe passes to each filter, and branch mains pass over its surface. The branch mains, which are perforated every four feet, become less and less in size. Each time the ejector goes off, 100 gallons of sewage is propelled through the perforated holes in the distributing pipes by means of the compressed air working the ejector. The sewage which is driven out through the holes impinges with a decreasing force against pieces of hoop iron, which are bent round the distributing pipe. This arrangement is well shown in the illustration.

The ejector takes less than ten seconds to drive out 100 gallons of sewage, and it is sprayed all over the surface of the filter. One filter has been constructed of coal, the other of destructor breeze. Both give excellent results. The filter has been at work about six years. The filter is under-drained by perforated pipes, and has no concrete floor or retaining walls. There are no ventilating pipes leading to the effluent drain, yet there is always a good current of air coming out at this pipe. This air must have passed through the filter along with the sewage. On August 14 and 15, 1901, I carefully measured the amount of air coming out at the effluent drain, and found that it varied considerably with the atmospheric conditions, and ranged in different readings from 80 to 140 cubic feet a minute, the latter obtained on a day when a strong wind was blowing. Sewage was being applied to the filter at the rate of 100 gallons a minute, while the air coming out of the effluent

drain was from 500 to 900 gallons per minute-that is to say, there was on this particular day with this filter five to nine times more air available for the oxygenation of the sewage than is the case with a contact bed.

Samples of the air coming out at the effluent drain of this filter have been carefully analyzed by me, and give the following results :

ANALYSIS OF AIR PASSING FROM EFFLUENT DRAIN OF CHESTERFIELD PERCOLATING FILTER, COMPARED WITH ATMOSPHERIC AIR.

As mentioned, at the time of the collection of these samples a strong wind was blowing, and the filter was in thorough action. For the sake of comparison, I have collected a sample of the air coming out from the filter on a still day after the filter had been allowed to remain resting for thirty-six hours. When the sewage was first applied, no air at all came out with the effluent. In about three hours 95.6 gallons of air came out with the effluent for every 100 gallons which were sprayed over the filter.

A sample of this air had the following percentage composition :

The

On the day of collection no wind at all was blowing. air was analyzed by direct absorption in a Hempel's gas pipette.

To ascertain the value of spraying the sewage over the filter, the sewage was collected throughout the day by means of a large glass funnel, as it fell in the form of spray, and the following analyses were made: