After filling the traps with water, one opens the stopcock between the trap to be tested and the pipe system, so as to allow air passing through the trap to break the partial vacuum therein. This air will follow the path of the least resistance. In the case of the S trap the special air vent pipe is provided between the trap seal and the waste system under the supposition that it will present an easier path for the air than the trap itself, in which the water seal might be expected to afford greater resistance to the air passage than the friction of the sides of the vent pipe.

The apparatus shows that this is not the case. The ordinary lavatory S trap is shown on the right-hand side of the apparatus vented at the crown with a "back vent" pipe the full size of the bore of the trap, and of considerable length, but having openings at different points provided with stop valves to show the effect of longer or shorter vent pipes in plumbing work. The longer the vent pipe and the greater the number of bends in it, the greater the obstruction by friction to the passage of the air through it, and the lower its efficiency as a means of protecting the trap seal from siphonage.

It is evident, therefore, that the degree of vacuum in the pipe system required to break the seal of a vented S trap is inversely proportional to the length and diameter of the vent pipe, and our apparatus is designed to show the effect of any possible degree of vacuum on any possible length and size of vent pipe, and it will again be seen on this apparatus that the seal of an S trap can be broken by a vacuum considerably below that which may be obtained in ordinary plumbing practice, even when the trap has a perfectly clean vent pipe of moderate length, provided a few bends are introduced in running the vent pipe.

If the friction is increased by more or less clogging the siphonage is by so much easier.

The only weak point in the employment of the ordinary unvented "pot" or "drum" trap of large size as a substitute for the system of back venting is that such a trap is not self-scouring.

This weak point is referred to by advocates of back venting as sufficiently important to justify the condemnation of the whole system of unvented trapping.

We have explained the fallacy of this argument by showing that the vent pipe is still less self-scouring, and that, once clogged, it can not be so easily cleansed. . Nevertheless, it must be admitted that the cesspool element in all reservoir traps is a very undesirable feature, and that not the least of its objections is the fact that it furnishes the opposition with an argument, however inadequate, against the use of the simpler system of plumbing.

Moreover, the use of cesspools in a system of house drainage is contrary to the first principles of sanitary engineering, and an evil which may be of comparatively slight importance when confined to the small proportions of a single fixture trap, becomes of very great importance when multiplied by thousands or hundreds of thousands in the complete drainage system of an entire city or town.

It is also perfectly true that the clogging up of the ordinary drum trap converts it in time into an S trap, and completely deprives it of its anti-siphon feature. So long as it is allowed to remain clogged, a system based upon the use of such traps loses its efficiency and, however far superior it may be to the "back vent" system, it can not yet lay claim to permanent efficiency with automatic action.

Hence it is all-important to know if the cesspool element of the common anti-siphon trap can be eliminated and the self-scouring property of the simple S trap or of a straight piece of smooth pipe of equal area can be combined with the anti-siphon quality of the ordinary drum trap.

For over a quarter of a century persistent and untiring efforts have been made to attain this combination, until at last it became generally believed that the two requirements were really antagonistic to one another, and that the solution, therefore, was as impossible as the discovery of flexible glass or transparent rubber, a large body of water and a deep seal being assumed to be necessary to withstand siphonage and the exact reverse being needed to produce the maximum of self-scouring effect.

It was found that resistance to siphonage increased with the increase of water head in the seal of the trap, and it is sufficiently evident that even in an S trap a seal deep enough to balance the weight of the atmosphere, say 32 feet, would be able to resist any siphoning action which could be brought to bear upon it, since Nature's abhorrence of a vacuum ceases under a water pressure of 32 feet. Accordingly, the idea has prevailed that depth of seal is an essential quality in anti-siphon traps, and nearly all such traps have been and are being constructed on this assumption.

It has been clearly shown that this idea is fallacious, and that exactly the reverse is essential to meet all the requirements of the problem.

It does not follow that because a depth of seal of 32 feet is capable of resisting completely siphoning action, there is no other equally efficient manner of accomplishing the same result without the objections attending the perpendicular form.

These objections are, first, the impossibility in practice of finding sufficient room under a fixture to give an unventec S trap the depth of seal required for adequate protection against the strongest siphonage encountered in plumbing practice. Four or five feet of such seal would be essential in an unvented S trap for entire safety. I have broken the seal of such a trap nearly six feet deep on different forms of apparatus.

Secondly, the resistance to the outflow and consequently the lowering of the scouring effect of water through the use of so deep a trap, and, finally, the prohibitive cost of its manufacture and installment.

Fortunately, there is another method of overcoming siphoning action besides that of opposing direct resistance thereto, namely, the diametrically opposite one of reducing the resistance to. a minimum. This is accomplished by forming the shallowest possible seal, but constructing the trap in such a manner that the seal shall simply move to one side for a moment and allow the air of siphonage to pass, and then return to its place. It is an application of the principle which gives success to the reed in competition with the oak. It bends to the storm and rises again when its fury has passed, while the oak is shattered by its opposition.

Nor does it follow, fortunately, that because a seal is very shallow it must on that account be a correspondingly weak one. As I have shown, a trap may be constructed in such a manner that the shallowest possible seal may have a strength so great as to be absolutely irresistible. A seal a quarter of an inch deep may be made in such a way as to withstand indefinitely the severest test of siphonage, back pressure, evaporation and all other adverse influences to be encountered in plumbing. It will withstand unvented a strain of siphonage powerful enough to destroy the seal of a fully vented S trap or of a drum trap of the largest diameter.

In my various experiments on the behavior of different sizes and forms of drum traps when subjected to siphoning action, I have found their resistance to be in direct proportion to their diameter rather than to their depth of seal. Indeed, the latter has, within the narrow limits practicable in construction, no appreciable influence whatever.