kinds of action the sewage is first led into a large tank, vaulted so as to exclude light and air, and is allowed to remain in this tank till putrefaction has taken place, when the sewage is passed upon the filter-beds which contain, as all filter-beds do, the nitrifying organisms. This method of treatment has been introduced into Exeter, England, and at present about 70,000 gallons are being treated each day.

The plant consists of the so-called septic tank, which is an underground tank built of cement concrete, sixty-four feet long, eighteen feet wide, and of an average depth of seven feet, having a capacity of about 53,000 gallons and of five filters made of coke breeze and furnace clinkers, about five feet in depth and covering all together an area of 400 square yards, having a capacity of 695 cubic yards. The crude sewage as it arrives at the plant passes through a railing to prevent large objects from entering the tank, while all small particles and solids in suspension pass freely to a grit chamber which is divided in two, each part having an inlet into the tank. The inlets are close to the bottom of the tank and the aperture of the inlet pipes is smaller on the tank side than on the sewer side, so that the sewage enters the tank with considerable force. The outlets are also beneath the surface of the water so as not to admit air or light, and so arranged that the water at the middle depth alone escapes. They are gauged so that the rate of flow may be measured; usually it is about 3,000 gallons per hour. The sewage from the septic tank is discharged simultaneously on two of the bacteria filters arranged according to Dibden's method. When these two filters are full they are emptied of their contents, and, while being emptied, the sewage from the tank passes on to others. The fifth filter remains idle one week, when it takes the place of the one that has longest been in use. Thus each filter remains idle one week out of every four. The filling and emptying of these filters is done automatically by an ingenious arrangement patented by Mr. Cameron and is said to be most satisfactory.

The action inside the tank is essentially of a putrefactive nature, no nitrates being formed. The organic matter is decomposed, or rather broken up into simpler forms, a large amount being rendered soluble, while at the same time ammonia and, I believe, free nitrogen are being formed. According to analyses

made by Dibden, the amount of oxidizable organic matter in solution is reduced about 30.8 per cent., the free ammonia about 26.9 per cent., the albuminoid ammonia about 17 per cent., and the suspended solids about 55 per cent. Pearmain and Moor, who have made a report on the process, state that there is no accumulation of sludge in the tank beyond a small amount of thin black sediment, which they report is so slight that a year's accumulation would scarcely be worth the trouble of removing. F. J. Commin, in a report on the process published in the Lancet for December, 1896, says that the deposit is very fine and inorganic, and that in a small tank after seven months' continual working and after quite 2,000,000 gallons of sewage had passed through the tank, the deposit was less than four inches over a surface of twenty-four feet by nine feet.

On top of the liquid in the tank there is a layer of flocculent matter from two to two and one-half inches in thickness, and from all accounts this seems to have been formed during the first few weeks that the tank was used and not to have increased much in thickness since that time. It appears to be composed of organic matter, formed I believe from the suspended matter in the sewage, and from all the information I have been able to obtain, it seems as though the organic matter in this flocculent layer was at first partially decomposed by the putrefactive organisms. Portions of it then sank to the bottom of the tank, where further action took place. Bubbles of gas collected around the fragments that had been carried down, causing them to rise to the surface and this process went on till the residue that remained at the bottom contained very little organic decomposable matter.

This action reminds one of what takes place in small streams into which sewage is emptied. I have often seen in such streams cakes of at least one-half yard in circumference rise from the mud at the bottom, and float down with the current.

According to Dibden the liquid that comes from the tank is of a brownish yellow color, offensive, and contains 2.73 grains per gallon of free ammonia, o.175 grain of albuminoid ammonia, and the amount of oxygen absorbed from potassium permanganate in four hours is 1.405 grains per gallon. It contains no nitrates nor nitrites. The original sewage contained at the

same time 3.778 grains free ammonia, o.212 grain albuminoid ammonia, and the oxygen absorbed from the potassium permanganate was 2.028 grains per gallon. This liquid is passed upon the filters as has been described and the effluent from these filters contains 1.705 grains free ammonia, o.078 grain albuminoid ammonia, 0.253 grain nitrogen as nitrites, 0.353 grain nitrogen as nitrates, no suspended matter, and the absorbed oxygen from the potassium permanganate in four hours is only 0.388 grain per gallon.

The great advantage of the septic tank process, if it does what it appears to do, is that so large an amount of suspended matter is removed from the sewage that there will be very much less trouble with the clogging up of the surface of filter-beds in winter, and consequently an area that is large enough for the purification in summer will be more nearly, or possibly quite, sufficient for the work during the winter months.

It is also claimed that a large amount of the organic matter in solution is removed in the septic tank and if this is so, which appears probable from the analyses that have been made, it may not be too much to hope that future developments in this direction taken in connection with the using of the cubical capacity of a bacteria filter, may so reduce the area required for purification, that filter-beds may, without too great expense, be protected from snow and ice.

WORCESTER POLYTECHNIC INSTITUTE,

WORCESTER, MASS.

INVESTIGATION OF THE THEORY OF SOLUBILITY
EFFECT IN THE CASE OF TRI-IONIC SALTS.

By ARTHUR A. NOYES AND E. HAROLD WOODWORTH.
Received January 20, 1898.

HE theory of the effect of salts on the solubility of one an

TH

other has, in the case of di-ionic salts having one ion in common, been quite thoroughly tested and confirmed.' The solubility of tri-ionic salts in the presence of other salts, however, has been much less investigated. The theory of the phenomenon has, to be sure, been already developed, and has been partially tested by experiments with lead chloride in the presence of other salts.* 1 Compare especially Noyes and Abbot: Ztschr. phys. Chem., 16, 125.

2 Noyes: Ibid, 9, 626.

There was found, however, to be only an approximate agreement between the theory and the facts, probably because the dissociation-values involved are uncertain. Moreover, the theory could not be tested in the case where a salt with a common bivalent ion was added, probably owing to the fact that a double salt was formed. A further investigation of the subject seems therefore to be desirable.

To this end we have determined the solubility of lead iodide in pure water, and in potassium iodide and lead nitrate solutions of varying strengths. We made use of lead iodide because of its slight solubility (one molecular weight in about 600 liters); for, in such dilute solutions, the influence of the two substances on the dissociation of one another is hardly appreciable, and the tendency to the formation of double salts is ordinarily very slight,-two phenomena which often have a disturbing influence in concentrated solutions. In order to determine the solubility, we have measured the electrical conductivity of the saturated solutions, and subtracted from it the conductivity of the water or of the solutions before treating with lead iodide. This method has the advantage over the analytical determination, of greater convenience; and in this case it is especially well adapted, as it furnishes directly a knowledge of the concentration of the ions, so that it is not necessary to consider dissociation-values.

Two samples of each of the three salts were prepared by different methods. These samples were in all cases shown to be exactly alike as far as the conductivity of their solutions was concerned, which indicates that the substances were, in all probability, pure. One sample of the potassium iodide was prepared by treating the commercial, chemically pure salt with alcohol until it was about one-half dissolved, and by subsequent crystallization from this solution by evaporation. The other sample was obtained by crystallizing from water, the residue left undissolved by the alcohol. The lead nitrate was in one case obtained by crystallization of a commercial sample from water; in the other, by precipitation with nitric acid followed by crystallization from water. One sample of lead iodide was prepared by metathesis from lead acetate and potassium iodide; the other, by dissolving a commercial preparation in a strong potassium iodide solution, and then diluting with water. In

both cases the salt was purified by crystallizing twice from water. The solubility determinations were carried out as follows: Small forty-cc. glass-stoppered bottles were charged with an excess of the solid lead iodide and with pure water or with solutions of lead nitrate or potassium iodide; the stoppers were coated with paraffin, and the bottles rotated for five hours in a thermostat by means of a previously described apparatus.' The solutions were then allowed to settle for a short time, and blown out by means of a wash-bottle arrangement into a resistance cell of the Arrhenius type; and the conductivity was measured in the usual way. All the solubility determinations and conductivity measurements were carried out at 25°. In all cases, duplicate determinations of the solubility were made in such a way that the condition of saturation was approached from both sides-that of supersaturation and that of undersaturation. Moreover, in order to make certain of the purity of the lead iodide, not only were the two different samples used, but each sample was also treated with successive amounts of water, and the conductivity of the corresponding solutions determined. A complete agreement was found to exist.

The following tables contain the specific conductivities (multiplied by 10) expressed in Siemens' units. In all cases the conductivity of the water used (which was equal to 9 X 10) has been subtracted.

CONDUCTIVITY OF WATER SATURATED WITH LEAD IODIDE.