To obtain these results, without internal complication or external aid, is only possible by taking full advantage of the various laws which govern the action of fluids in plumbing. The difference in the specific gravity of air and water, and the consequent difference of momentum of the two fluids under equal rapidity of motion, and the relative attractive and cohesive forces of the particles of the two fluids, give us reliable means of separating the air from the water in their passage along the inner walls of the trap as simply and unfailingly as chaff is separated from the grain in the winnowing machine.

I have given as the second of the three methods tried for protecting the seals of traps from loss by siphonage, the use of a large unventilated "pot" or "reservoir" trap. I have shown that a small pot trap will not resist siphonage, and that none which is less than eight inches in diameter can be relied upon in all cases; that a five-inch pot trap might sometimes be siphoned out by discharges from fixtures under conditions which may occur in practice; that a four-inch pot trap siphons out much easier; that an ordinary three-inch trap has very little resisting power, and that two and a half inch and two-inch traps are altogether useless, and but little more than S traps.

But unfortunately the larger forms of pot traps are, as has been said, not self cleaning. They are cesspools and violate one of our main principles of plumbing which prohibits the retention of decomposing waste matter anywhere within the system. They are also very bulky and expensive in use of material.

In order to better study these movements, we have had a large number of traps constructed in whole and in part of glass.

The initial cut, Fig. 138, shows one of our experimental glass traps in perspective, so constructed. The body is thirteen inches square and an inch and a half deep. The inlet and outlet arms are made of inch and a half pipe. The inlet end descends below the bottom of the drum instead of entering the side, as is customary.

Under siphoning action the seal standing in the inlet bend rises into the drum and simply stands one side while air passes through the trap from the fixture above the water in the drum, as shown by the arrows in the drawing, and breaks the partial vacuum in the soil pipe. The storm having blown over, the water seal quietly returns from the drum or reservoir chamber into the bend and restores the original conditions as a reed rises after the fury of a hurricane has passed.

Thus the trap becomes its own back vent pipe, a back vent pipe which has no inaccessible body waiting for mischief, which provides entire security, and yet which adds absolutely nothing to the expense.

As will be seen, a small portion of the water in the trap will be thrown out at the first application of the siphoning strain, but as soon as the level of the water in the reservoir chamber has been lowered a little below the overflow point, far enough to provide for the wave action produced by the air blast, no further loss of water can be occasioned even by the severest strain that can be brought to bear upon it. This trap was found capable of withstanding a strain severe enough to empty an S trap fully vented under the most favorable conditions with a new clean vent the size of the bore of the trap and only fifteen feet long. It perfectly illustrates the fact that the principle of resistance to siphonage lies not in depth but in breadth of seal. The maximum of strength comes with the maximum of horizontal dimension, but with a minimum of height.

The trap is, however, still open to the objection that it is not self-scouring. The sediment chamber is not so large as it would be in a deeper drum trap. The cesspool feature has been eliminated only in one of its dimensions.

Fig. 193 shows the manner in which a pot trap of this form, though absolutely antisiphonic, could clog with grease under a sink.

The third method of obtaining the desired security is, as stated, to obtain some form of trap which shall be both antisiphonic and self scouring at the same time.

In our experiments with pot traps of various diameters, from eight inches down to two inches, we have found that with traps of equal depth their resistance to siphoning action very rapidly increased with the increase of their diameter; that with traps of equal diameter their resistance to capillary action increased with their depth; that resistance to back pressure increased with the increase of water capacity of the trap and with its depth.below the fixture it serves; and that resistance to fouling action and clogging increased as the sectional area of the body of the trap approached that of its inlet and outlet arms; and that, finally, resistance to evaporation increased with the increase of water capacity of the trap and of its distance from air currents. I have moreover lately found that a shallow seal trap may be designed in such a manner as to protect the seal of a water closet trap from siphonage, as will hereafter be shown.

From this it would appear, at first thought, that to obtain a trap capable of affording the maximum resistance to all these adverse influences at once and under all conditions would be impossible, because the desiderata above enumerated seem to be in direct conflict with one another, a large diameter being needed to resist siphonage and a small one to resist clogging, while evaporation and capillary action seem to demand a deep seal and thorough scouring, and water closet trap protection a shallow one. But a closer investigation will make clear that these disiderata are not necessarily incompatible with one another, as the following experiments and reasoning will show.

The trap must be so formed, in the first place, that its sectional area shall in no place exceed the area of the fixture waste pipe which it serves, because otherwise it would not possess the maximum of self scouring power.