which he certainly had felt unable to explain at the time. The Mr. Corbett. experimental work was started about 1898. Twelve years ago he designed a scheme for the Salford works which the Committee had not yet allowed him to complete, though two-thirds of it had been executed, and some of it had been working for 5 years past. In the experimental works, which were in use for a number of years, the process was the following. First, there was thorough chemical precipitation. Various processes were tried at great expense, and the tank-effluent at that time was very good. Now a minimum of chemicals was used, simply to counteract the worksrefuse in the sewage. The second part-and to his mind a very important part of the process, was the use of what he called "roughing-filters," which were really strainers of sand and gravel for eliminating the particles which had escaped chemical precipitation. In the experiments the roughing-filters were washed out almost every day, and a remarkable amount of filth was taken from them, even after very careful chemical precipitation. Then the doubly-prepared sewage was put on to the bacterial beds. The contact system was not used, the system being that of trickling beds. The results were so remarkable as to be quite puzzling. For a number of years about a dozen separate filters were used, the working of each being carefully recorded and analyses taken, and the filters were used at a constant rate of 500 gallons per square yard, or about 2 million gallons per acre per day. After a number of years the Committee determined that, instead of the filters resting 12 hours and working 12 hours, they should be worked night and day. Two filters were actually so worked for about a year at the rate of 1,000 gallons per square yard, or 5 million gallons per acre, with thoroughly successful results. For a few weeks they were even worked up to 1,400 gallons per square yard, or 7 million gallons per acre per day. The results were always good. The standard aimed at required less than 1 grain of oxygen per gallon in a 4-hour permanganate test, and that was scarcely ever exceeded; it was generally about 0.6 grain per gallon. The actual tanks and bacterial beds had been completed, and were working thoroughly well. A chemical process was also used. He had not yet been allowed to complete the intermediate process of the roughing-filters, and it was not possible to get nearly as much through the filter-beds as should be got through with roughing-filters washed every day. The Authors' investigations indicated that probably what was being done unknowingly at Salford was to intercept colloidal matter which might have injured the results in the bacterial beds. At any rate, there was the broad fact, verified by many years of experiment, that, with good

н 2

Mr. Corbett. roughing-filters, it was possible to produce a good effluent with twice as rapid a flow through the bacterial beds as was obtained at present. Although now merely rough gravel and no sand was being used, with no means of cleaning, and the roughing-filters were only cleansed once in a while by hand-labour, still the bacterial beds had not choked seriously. It appeared that even a crude roughing-filter was of some value. He had been a little surprised to hear the Authors, after stating that it was desirable to expose as much surface as possible to attract the colloidal matter, say that they used material up to 6 inches in diameter. In preparing his own roughing-filters he used material ranging between and 1 inch in diameter, and it seemed to him that the Authors would get better results from material inch or inch in diameter. He hoped to be able soon to give results on the full flow of the Salford sewage, but at present only about one-half of the daily flow of 12 million gallons was being filtered. As soon as the work was completed he would be able to tell more certainly whether the results actually tallied, or were comparable, with those given in the Paper.

Mr. Taylor.

Mr. G. MIDGLEY TAYLOR thought everyone would agree with Mr. Lloyd-Davies that the ordinary kind of storm-overflow, with free sewer and the sill placed at the level attained by six times the dryweather flow, was antiquated and did not fulfil the requirements; and the same might be said of the arrangement in which a portion of the outlet-sewer was masked and the overflow came into operation with the aperture under a small static head. With regard to Mr. Lloyd-Davies's recommendation of the separating-plate type of stormoverflow, his statement that the velocity of the sewage entering the overflow-chamber was reduced by the diminished hydraulic radius of the U-shaped channel with the plate on the top was correct, but he did not take into account the small static head which he mentioned in connection with the second type of storm-overflow. If he did so, he would find that his design of storm-overflow did not fulfil the conditions any better than the other. Mr. Taylor had recently designed for the main drainage of the City of Glasgow a stormoverflow of somewhat large dimensions, and, he believed, of entirely novel form. The problem submitted to him was the following:-In an 8-foot barrel sewer (Figs. 1), which might run under a slight surcharge, and would deliver as much as 22,000 cubic feet per minute to the overflow-chamber, it was desired to send forward only 7,600 cubic feet per minute; so that 14,400 cubic feet per minute had to be discharged by the overflow-weir. Having regard to the limited dimensions of the overflow-weir, which was about 45 feet long, it had become apparent that the depth over the weir would be so considerable

a weir, thus giving an effective length of 90 feet. The sewage coming from the main sewer passed into a U-shaped conduit

that it was desirable to construct the sewer in the form of a trough Mr. Taylor. through the chamber, so that both sides of it could be utilized as

to the conclusion that the main object to be aimed at was to have a gradually decreasing hydraulic radius in the channel so as to check the

Mr. Taylor. (Figs. 2) designed to check as far as possible the excessive velocity due to the surcharge. He and his brother, Mr. Brough Taylor, had come

CROSS SECTIONS OF STORM-OVERFLOW: MAIN DRAINAGE OF GLASGOW.

velocity and at the same time to discharge the liquid over the weirs. Mr. Taylor. The liquid coming in at one end was discharged on both sides, that which fell over on the one side passing under the sewer through arches and down the storm-overflow to the River Clyde, leaving a half-full sewer discharging towards the purification-works.

It was a curious thing that the first Paper should deal with the means of eliminating storm-water from sewerage systems, and that the second Paper should start with a recommendation of the separate system of sewerage. The Authors of the second Paper stated that since 1885 drainage by a common outfall-sewer had been deemed inadmissible, which he thought was an unfortunate statement to be introduced into such a Paper, as it was certainly open to criticism. With regard to colloidal matter in sewage, the Authors stated that hitherto it had been assumed that if the sewage was freed from its suspended matters in the first instance, then the filtration-area would be effectually protected from the intrusion of depositable matters. That certainly had not been assumed by himself, nor did he think it had been by any sanitary engineer who had had to deal with bacteriabeds. The sewage of Shepton Mallet contained more colloidal matters than any other sewage with which he was acquainted, and consisted of half brewery-refuse and half domestic sewerage. The brewers treated their refuse chemically and discharged it into the sewers, and on arriving at the sewage-works the mixture was again chemically treated with a little alumino-ferric and then with lime. It was next passed into settling-tanks, and the resultant effluent was a viscous, muddy-brown, but clear liquid, which if shaken in a bottle would carry a head like strong beer for a long time. In the ordinary course of events the liquid was further purified by being passed over land, but as at all times the effluent was not entirely satisfactory, a contactbed had been tried. It had then been found that the liquid coming from the contact-bed was as brilliant as ordinary drinking-water, but unfortunately in the process the contact-bed had become choked with colloidal matter in about 3 months, and the bed had become practically solid. He had determined therefore to try the effect of a streaming-filter and the moment the tank effluent was turned upon the filter it became practically as clear as drinking-water, and had no unpleasant smell. When the streaming bed had been in work for about 10 days the effluent had looked as though a handful of coal-dust had been sprinkled in a glassful; the solid matter had precipitated very rapidly, but unfortunately, being in a putrescent condition, it had had to be treated as sludge and separated from the liquid. He thought he had said sufficient to show that colloidal matters had not been lost sight of by sanitary engineers. The hydrolysing-chamber and