ground becomes frozen, the sewage usually freezes also if flowed over a flat surface in a thin stream. If, however, the land be deeply furrowed, there is little danger of the sewage freezing. If the land is only slightly porous, flooding to a depth of a foot or two will give satisfactory results. The sewage should be kept as warm as possible before discharging onto beds. There is little bacterial action when the temperature of the sewage is below 40 degrees; the temperature most favorable for rapid oxidation appearing to be 90 degrees; at about 130 degrees it entirely ceases.

Worms and burrowing animals occasionally give trouble by opening passages in the soil by which unpurified sewage reaches the drains. These have been driven out by flooding the land once or twice with very strong or septic sewage.

The sludge from the settling tanks is generally pumped or flowed upon beds set apart for this purpose, and each application is raked off after it has dried, and the deposit is left piled upon the surface to be burned. In a few plants the sludge is taken by farmers for fertilizer.

The cost of land for irrigation or filtration plants will of course vary with every city. To a certain extent the cost of preparing the plant also will vary, depending upon the character of the soil and the nature of its surface. Of several plants in this country, the cost of construction of those using natural soil was from $700 to $1500 per acre; and the cost of those constructed artificially of sand and gravel was from $1500 to $6000 per acre. Operating these plants cost from $75 to $100 per acre per year, or from $5.50 to $9.00 per million gallons.

ART. 80. CONTACT FILTERS. SLATE BEDS

An intermittent filter produces the purest effluent practically obtainable from sewage. But the rates are low; and in some cases a less pure effluent would be satisfactory if less area of land could be used. This is found to be impracticable with a finegrain filter, but should theoretically be with a coarse-grain one. The latter, however, presents the practical difficulty of obtain

ing a uniform distribution of the sewage throughout the filter. If flowed on, as in the case of a fine-grain filter, the sewage passes through a small section only, near the point of application. To meet this difficulty, the contact filter was devised. In this, the sewage is allowed to fill slowly a bed composed of stones (generally of a size varying from pea to walnut), to stand in it while the suspended matter settles onto the stones or collects on them by surface adhesion, and is then withdrawn slowly; after which the bed is allowed to remain empty for a few hours to become reaërated and permit oxidation to take place. In many cases two hours is allowed for each step, or eight hours for a cycle.

The theory of action of these filters is as follows: "When the effluent flows from a filter, air is drawn into the filter again and fills the open space. Consequently a partial oxidation of the organic matter left within the filtering material proceeds until this oxygen is exhausted, when the open space is completely filled with the chief products of this oxidation, namely, carbonic acid gas, marsh-gas, nitrogen of the air primarily present and nitrogen liberated during decomposition, and the filter will remain with its open space filled with these gases until they are removed by the introduction of sewage or air. This condition reached, the activity of the oxidizing and nitrifying bacteria within the filter ceases and anaerobic actions begin, which change a considerable portion of the organic matter adhering to the filtering material into forms easily soluble and oxidized by the air introduced when the filter is again flooded." (Mass. State Board of Health, 1899.) If these filters are used in pairs, the effluent from the "first-contact filter" passing to the "secondcontact filter," the action in the latter becomes almost wholly aerobic.

A contact filter consists of a pit, generally about 4 to 8 feet deep. The pits have generally been made watertight, but this does not seem to be essential; and experimental ones at Manchester were simply excavated from the soil, with side slopes of 2 to 1. On the bottom of the pit is laid a series of drains leading to a main cutlet pipe, which is provided with a valve for regulating the flow of sewage from the filter. The pit is then filled

with coke, coal, slag, cinders, gravel, burnt clay, glass, or other clean, insoluble material of fairly uniform size. Coke breeze gives excellent results, although it is liable to slow disintegration. The Manchester experts obtained their best results from clinkers passing through 14-inch mesh and rejected by -inch; and this material is recommended by the Massachusetts Board of Health. Both of these bodies of investigators found that the contact beds had at first a water capacity of about 50 per cent, but that this

was quickly reduced to about 33 per cent, at which it remained constant; the reduction being due partly to the growth of bacterial jelly on the surfaces of the filter material, partly to chaff, straw, and wood and cloth fibers. To prevent the filling of the filter by sand or other solid mineral matter, a pit or catch-basin should be placed above the filter, through which the sewage should flow at such velocity as to carry on all but heavy insoluble

matter.

As already stated, the operation of a contact filter consists in filling the filter, allowing it to remain full for a fixed time, emptying, and allowing it to stand empty; two hours being

allowed for each operation in many cases. It was found at Lawrence that if the sewage stood but two hours in a singlecontact bed which was filled once daily, the action during this time was anaerobic only, the aerobic action taking place while the tank stood empty. The rests between doses should not be long enough to permit the bacteria to die from lack of pabulum, but these should be preserved in the filter to work over successive doses. For this reason also the sewage should not be allowed to enter or leave the bed with so great velocity as to wash the bacteria out of the filter.

FIG. 72.-CONTACT BEDS WITH TILE PIPE DISTRIBUTORS AND HAND GATES.

In order to secure this rotation, it is necessary to hold the sewage in a tank between dosing periods, or else to have four or more beds to be closed alternately. In either case, the flow of sewage to the beds and of effluent from them must be controlled at intervals. This may be done by hand, or by automatic dosing appliances, of which there are several on the market.

If a contact bed is filled three times a day, and its interstices have a volume one-third that of the entire filter, it is evident that the daily capacity of the filter is its cubical contents. A filter 5 feet deep could therefore treat 37 gallons per square foot per day. Allowing for walls or embankments between filters and occasional resting or cleaning of beds, it is thought that 25 gallons per square foot per day, or, say, 1,000,000 gallons per acre, can be purified. If double contact is employed, as it should generally be, double the area will be required; or 500,000 gallons

per acre per day can be rendered unputrescible; which was the conclusion reached by the Manchester Commission.

Double-contact filters, 6 feet deep, in London have removed practically all the suspended matter and 51 per cent of the dissolved putrescible organic matter, when receiving 600,000 gallons of crude sewage per acre per day. Dibdin in 1895 filtered through 3 feet of coke breeze the effluent from a lime-precipitation plant at the rate of 1,000,000 gallons per acre per day, the effluent from the contact filter containing 71 per cent less albuminoid ammonia and absorbing 77 per cent less oxygen than the precipitation effluent which was applied to it.

In tests at Columbus with both single and double contact, the effluents from the primary contact beds were putrescible for about one-third of the time; and those from the secondary contact filters were found putrescible about 25 per cent of the time; the rates being from 100 to 300 gallons per cubic yard per day, which was considerably reduced by periods of rest which were allowed at intervals. The tanks were 5 feet deep and the net rates of treatment varied from 0.5 to 2.38 millions of gallons per acre per day; averaging about 1 millions. No odor was noticed around the filters, and when the material was removed for cleaning the only odor noticed was that characteristic of garden soil. The percentage of suspended matter removed varied considerably, but averaged about 40 to 50 from crude sewage, and 60 to 70 from settled or septic sewage. Of the organic nitrogen the average removal was about 35 to 40 per cent. Of the bacteria the percentage of removal varied all the way from 0 to 60, averaging about 40. Of the applied nitrogen, there appeared in the nitrified form in the effluent of the primary filters from 4 to 11 per cent and in the effluents of the secondary filters from 17 to 21 per cent. An important feature was the uniformity of removal and the absence of any such unloading of stored material as is characteristic of sprinkling filters.

These filters had voids amounting to from 43.1 to 54 per cent of their volume at the beginning, which was reduced to from 31.9 to 45.8 at the end, these voids being somewhat greater than had been found in other cases. Both limestone and coke were found