Therefore the trap will resist siphonage as long as there is any water way at all left in the trap. Yet when the discharge is entirely stopped by sediment, or retarded to a point of inconvenience, it will, of course, announce itself and necessitate opening and cleansing.

The same advantage holds in the comparison of this trap with a pot trap or any other form of plumbers' trap constructed on the unscientific and faulty "vertical" principle, which, strangely enough, is the one on which plumbers' traps have always heretofore been erroneously designed.

As has already been explained in a former chapter we have subjected our traps to strains of various degrees of intensity, the severest being much stronger than any which could be encountered in plumbing practice, for the purpose not only of proving a degree of resistance beyond all possible question on the part of the antisiphon traps tested but also of permitting a more thorough comparison between the various forms of traps under consideration, and especially between unvented antisiphon traps and ordinary S and pot traps fully vented in accordance with the present plumbing laws.

It only remains to determine how far it is best to contract the horizontal dimensions of our trap in order to obtain on the whole in practice the most desirable results.

Our next experiments therefore were made to decide this question. Figs. 181 to 189 inclusive show the various sizes of spiral traps experimented upon arranged in the order of the tests.

The endurance of each of these traps is recorded in the Table I. The depth of seal in all was the same as in all the preceding horizontal traps; i. e., 1 inches. The most that could be forced out of the 13-inch spiral trap even after numerous successive repetitions of the ordeal was 3/8 inch in one set of experiments and 17-32 inch in another.

The 10-inch spiral (Figs. 180 and 181) lost only inch under the same tests. The 10-inch spiral trap lost 9-16 inch. The 8-inch trap lost 7-16 inch and the 7-inch lost 9-16 inch, all under the same tests.

Table III Experiments on Water Scour 191

Fig. 180.

Table III Experiments on Water Scour 192

Fig. 182.

Table III Experiments on Water Scour 193

Fig. 181.

Table III Experiments on Water Scour 194

Fig. 183.

Table III Experiments on Water Scour 195

Fig. 184.

Table III Experiments on Water Scour 196

Fig. 185.

Table III Experiments on Water Scour 197

Fig. 186.

Table III Experiments on Water Scour 198

Fig. 187.

Table III Experiments on Water Scour 199

Fig. 188

Experiments were also made with a 6-inch spiral trap, and this lost 1 1/8 inch after four of these severest strains in succession. These strains long continued would have ultimately broken the seal of so small a trap. But it withstood all other strains as shown, and proved itself capable of easily withstanding any strains of siphonage which can ever be encountered in actual plumbing practice.

The arrangement of partitions shown in Fig. 184 seemed to give results not appreciably different, so far as siphonage is concerned, from those of Figs. 180 and 182. But the sharp bends between the inlet and the outlet arms somewhat increased eddies and the friction in normal use and obstructed the free discharge of heavy substances in the waste water.

The small opening shown in Fig. 187 between the outlet pipe and that part of the spiral which is nearest to it produced a scarcely appreciable effect in the siphonage tests. It would, however, be objectionable as a cause of complication and possible obstruction and its use was abandoned.

In Figs. 186 and 185 corners were rounded off as indicated by the black places in the drawings. This reduced the resistance to siphonage by so small an amount that its advantages in facilitating scour much more than offset the loss. In Fig. 186 the bottom of the trap at the inner end of the spiral is curved gently upwards in order to do away with any sharp corners and barriers. This also improved the scouring properties of the trap without appreciable injury to its resistance to siphonage.

Before describing our final step it will be interesting to record certain curious facts noted in making our experiments on our horizontal traps not heretofore observed or recorded, so far as I am aware.

For the purpose of studying the movements of waste water through very large shallow traps we had the one we have shown in Figs. 189 and 195 constructed with a glass top, the length and breadth being 13 inches each and the depth 1 inches. The seal proper was, as in the other cases, only half an inch deep, and the water stood inch deep in the reservoir chamber when full up to the overflow, making a total seal of 1. inches under normal conditions.

The actual movement of the water in this trap, under both siphonage and friction tests, proved quite different from what might naturally be expected. One might suppose that under the pressure (or "suction" as it is popularly called) of a powerful siphoning action, air and water would be forced straight across the reservoir from inlet to outlet arm along the line of the least apparent resistance, in a straight and rapid current somewhat as shown in our figure, with return eddies on each side of the main current. It would also be natural to expect some such current to be formed when water was discharged through the trap with considerable force from a fixture connected up as shown in Fig. 190, where we have used a 12-gallon tank set 19 inches above the trap, to represent normal discharges from an ordinary bath tub.