the intermediate deposits nearly to the upper limit of the wave and leave them there.

The interval which should elapse between flushings will vary under different conditions. In sewers where there is a constant ample flow of water, where stoppages are few and due solely to accident or to design of ignorant or malicious persons, flushing need be resorted to only when such stoppages occur. If it is found from experience that stoppages are frequent or that there is a constant depositing of material in certain sewers, or if it is foreseen that this will occur, frequent flushings of these should be provided for. In the case of a dead end of a separate or combined sewer, or one which has but few house-connections made with it, the flushings should be done once in each twenty-four or at least fortyeight hours.

Both separate and combined systems have been built and satisfactorily maintained without flushing at any point oftener than two or three times a year. It is probable that this is possible only where there is considerable ground water entering the sewers at their upper ends, or where the dead ends occur only in thickly populated districts and on grades a little greater than the minimum herein advocated. There is too little definite information on this subject to justify a positive statement as to when, if ever, flushing at dead ends may be profitably omitted. It is advisable so to arrange every separate or combined sewer, where the conditions will be those given as favoring deposits, that it can be satisfactorily flushed.

Investigations made of the effect of flushing appear to show that 300 gallons is in most cases insufficient to properly flush an 8-inch pipe, at least 400 or 500 being necessary for effective scouring; also that the effect of such a quantity is felt for about 800 to 1000 feet.*

In flushing by hand, the sewer may be stopped at the downgrade side of a manhole or flush-tank and this be filled to the desired height with water or by allowing the sewage to accumulate in and above it, the gate, plug, or other stopper, is

See also Transactions Am. Soc. C. E., vol. XL, pp. 1-30.

then removed and the water allowed to enter the sewer under the head due to its height. Where outside water is used for flushing and is limited in quantity, another stopper may be placed at the upper orifice, in the case of manholes, to prevent a flow up the sewer, and be left in until the flushing is over. The stoppers are made of various forms and to act in various ways, and to close the whole sewer or only the lower half or two-thirds if the sewer be large.

Some have advocated admitting roof-water to separate sewers, claiming that it is beneficial in flushing the sewer. If roof-water is admitted to small sewers throughout their length, there is great probability of its gorging the pipes and backing up into connected basements and cellars. In several cities great damage has been caused in this way. If it is admitted only at and near the dead ends it usually will be advantageous, but it should not be thought to take the place of all other flushing, especially where daily flushing is desirable. The sewers are most likely to need flushing at dry seasons, and this must then be done by hand or otherwise, and there is a danger that the presence of these roof connections will give a false idea that the flushing requirements have been entirely met.

In England the separate system, when first constructed, was designed to admit to the separate sewers roof-water and drainage from yards, and this method is still followed there to a considerable extent. But since the danger is so imminent and the benefits contributed at such uncertain intervals, most American engineers do not advise the admission of roof-water to small sewers.

Sewers are sometimes flushed by connecting their upper ends with convenient streams, or artificial channels filled from such streams, the water being admitted periodically by gates.

Tides are sometimes made use of for this purpose, the water being allowed to rise in the sewer at high tide and being held there by gates until the low tide, when it is released. Ordinarily only the lower reach of the outlet sewer can be flushed thus. A better method in some cases is to hold the

water after high tide in a basin, from which it is rapidly discharged at low tide into the sewer to be flushed.

As in the case of Milwaukee, Bremen, and a few other large cities, the flushing-water may be pumped from a lake or river directly to the sewer. This is, of course, applicable within the limits of economy to very large sewers only, or to a system where a number of dead ends can be reached by a comparatively short line of force main.

The water for flushing is sometimes taken from the ocean or other body of salt water; but salt water mixed with sewage causes suspended matter to settle more readily and is corroding to any metal-work in the sewers. Hence its use is not advised by most authorities.

Automatic flush-tanks are in use in a large number of separate systems, but are seldom used for flushing combined or storm-water sewers, owing to the enormous quantities of water needed for that purpose. A great number of devices have been invented for flushing, but practically all of those now used are siphons in principle, so arranged that a tank in which they are set may fill gradually up to a certain point, when its contents are discharged rapidly into the sewer by siphon action. The tanks are made to contain at the time of discharge from 250 to 600 or even 1200 gallons for 6- to 10-inch pipe sewers. For larger sewers larger quantities are provided. The smaller quantities are of little use. No tank should discharge less than 250 gallons at a time into a 6-inch pipe, and correspondingly larger amounts into larger sewers. Five hundred to 800 gallons discharged into an 8-inch pipe once in twenty-four hours would be more beneficial than half of that amount at each of three or four discharges during the same time. The tanks should, of course, be watertight. They are usually built of brick or concrete (the latter often reinforced) plastered on the inside. Wood or steel could be used, but would not be so durable. They should be so built and arranged that the water may have the greatest permissible head above the sewer when discharging. (For details see Art. 42.)

The water may conveniently be admitted to the tank through

a stopcock connected with the street-main by a supply-pipe passing through the tank-wall. This cock is continually left sufficiently open to cause the tank to fill and discharge at desired intervals. If the water is inclined to be muddy at times, the use of too large a supply-pipe will result in the choking of it by sedimentation; and tuberculation or corrosion may stop it if a metal so affected by the water be used. To secure a velocity of flow through it of 2 feet per second to prevent deposits would require a diameter of less than half an inch if 800 gallons is fed to the tank each twenty-four hours. Some meet the difficulty by using a larger pipe with a stopcock on. the end which is kept nearly closed, but which is opened wide at intervals to flush out the pipe. Instead of a cock, a small orifice plate is sometimes placed at the end of the pipe, the size of the orifice determining the flow.

Where automatic flush-tanks are not used, some engineers have built into manholes at dead ends 2-inch to 4-inch pipes connected with adjacent water-mains and provided with gatevalves. This is probably the most convenient method of hand-flushing and the cheapest to operate. The cost at Mount Vernon, N. Y., averaged about $40 for installing each 4-inch branch and connection. In other cases the outlet in the bottom of the tank is provided with a plug or flap valve that is lifted by an attached chain when the tank is to be emptied; the tank being filled from the water main by a small pipe and stopcock, or by hose from a fire-hydrant.

Whether the tank be automatic or hand operated, the discharge pipe should be at least as large as the sewer; and it would be better to have it a few inches larger and bellmouthed at the end so as to secure the maximum obtainable velocity of flow in the sewer.

Provision for flushing large sewers, and sometimes small ones, may be made by building in any manhole a gate of some kind that closes the sewer opening and allows the sewage to collect behind it, which gate can then be opened by an operator on the ground above.

ART. 18. SEWER AIR

In every sewer there is a space above the sewage filled with air, and this air will generally be far from pure unless kept in motion and frequently renewed. The odor accompanying all sewage, even when there is no decomposition proceeding in the sewer, is communicated to this air, and some gases due to putrefaction may be given off. This air probably is seldom motionless. It is influenced by the sewage to move down the sewer; it is warmer in winter and often cooler in summer than the outside air, which condition occasions motion when there is communication between the two; it is driven out of or along the sewer by sudden inflows of sewage from house-connections or branches and sucked in by decrease in the volume of flow; near the outlet, the direction and force of the wind affect it, driving it up the sewer or sucking it out; last, and most important, it passes into empty or partly empty house-connections and into proximity to, if not into the air of. connected residences. There is no "sewer-gas" which is deadly to human life, but air that has been confined in contact with decomposing sewage may become charged with sulphuretted hydrogen, ammonia and other gases that are objectionable to the smell and may be more or less detrimental to health, although the weight of evidence appears to indicate that no pathogenic bacteria are ever carried into house plumbing pipes by sewer air. In fact, the air in sewers where there is ordinarily good ventilation is generally found to be more pure than that in a theater, church or other room well filled with people.

The most serious pollution of sewer air is that due to gasoline, and next to this, illuminating gas. Gasoline is discharged into sewers by dry cleaning establishments and garages, and there vaporizes. When the vapor is mixed with a certain percentage of air the combination is highly explosive and may be fired by a spark from a horse's hoof on a manhole cover, a smoker's match, or possibly by spontaneous combustion. Hundreds of thousands of dollars' damage and several deaths have been caused by such explosions. The only preventive