What Siphonage Is and Why Only a Vent Stops It

Why this matters

A siphon does not need suction from anywhere. It needs a full pipe running downhill and no way for air to get in behind it, and a drainage system builds both of those by accident every time a fixture empties. That is why a lavatory that worked for fifteen years starts gulping the week after somebody moved it six feet along a wall. Nothing broke. The geometry that used to keep air within reach of the trap stopped being true. This article is about that race between water and air, the one measurement that decides who wins it, and why every fix that is not an air path only postpones the loss.

What a siphon needs

Two conditions, both of which have to hold at the same time.

A continuous column of water filling the bore, running downhill. A pipe with an air core in it cannot siphon, because the air core is already the vent. The column only has to be a few inches long to start the process, and once it starts, gravity on the downstream side keeps pulling.

No air able to reach the water's upstream face except through the trap. If air can get to the back of that column from anywhere else, atmospheric pressure fills the gap and the column separates. The siphon ends the instant that happens, which is measured in a fraction of a second.

Put those together and the mechanism is a race. Water is trying to keep the bore full and pull the seal over the dip. Air is trying to reach the outlet side of the trap. The winner is decided by which path is shorter, and a vent is nothing more cunning than a deliberately short air path installed where the race would otherwise be lost.

Self-siphonage and induced siphonage

Self-siphonage is the fixture emptying itself. A lavatory or a small tub dumps its contents fast enough to fill its own trap arm, the arm becomes the siphon leg, and the last of the discharge takes the seal with it. The tell is a gulp right at the end of the drain-down, on the fixture's own use, with nothing else running.

Induced siphonage is somebody else's water. Flow past a branch connection, most often down a stack, drags air with it and drops the pressure at that connection. The trap on that branch sees the outlet side go negative and gives up its seal. The tell is a gurgle at one fixture when a different one runs, and the seal-loss article covers how to separate that from a positive-pressure event, which sounds similar to a customer and is the opposite fault.

Both are negative-pressure events, and both are fixed by the same class of remedy: get air to the trap's outlet side sooner than the water can get the seal over the dip.

The trap arm is the whole geometry

The trap arm is the run from the trap's crown weir to the vent connection. Everything in self-siphonage is decided there.

The governing rule in the model plumbing codes, as adopted and amended by your jurisdiction, is that the total fall of the trap arm over its developed length must not exceed one pipe diameter. Developed length means centerline distance along the pipe, following the fittings, not the straight-line distance across the floor, and on an arm with an offset in it those two numbers are not close.

The reason is direct. As long as the fall stays within one diameter, the crown weir stays above the invert of the vent connection, and there is a continuous air path over the top of the flowing water from the vent back to the trap. Let the fall exceed one diameter and the far end of the arm can run full during a discharge. A full-bore section with a vent on the wrong side of it is a siphon leg, and the trap is the only thing left with air behind it.

That rule also explains where the maximum trap arm lengths in the code tables come from. Divide the pipe diameter by the required slope and you get the length at which the fall reaches exactly one diameter. At a quarter inch per foot, a 1-1/2 inch arm runs out at 6.0 feet, a 2 inch arm at 8.0 feet, a 3 inch arm at 12.0 feet. Run the check on the NOMINAL size, which is the basis the published trap-arm table reconciles to. A slightly larger internal diameter is not margin you get to spend. The published table is what governs on a permit; this derivation is what lets you see at a glance whether a proposed layout is anywhere near the edge.

Slope itself is owned by the slope and venting article and is not re-derived here. What matters for siphonage is that the required minimum slope is a floor set by the need to keep solids moving, so you cannot buy trap arm length by flattening the pipe.

Why depth and size delay rather than stop

Both of the usual instincts are real effects pointing in the useful direction, and neither one ends the mechanism.

A deeper seal gives the siphon more water to move before air breaks through, so a fixture that used to lose the whole seal on every use might now lose most of it. The siphon does not stop when the water gets deeper. It stops when air arrives. Depth buys events, not immunity, and on a fixture losing more than a full seal per event it buys nothing at all.

A larger trap body holds more water without changing the dip-to-weir height, which is the only dimension that matters, so it changes almost nothing. A larger trap ARM is different and genuinely helps, because it raises the fall allowance and it makes the arm harder to fill in the first place.

Slower flow into the fixture reduces how completely the arm fills, which is why the same layout can be fine on a lavatory and marginal on a tub. It is a real variable and a terrible fix, because it depends on the customer.

The only class of fix that ends the mechanism is admitting air. A conventional vent connected within the arm limit does that. An air admittance valve is also an air path and will break a self-siphon, but it is a one-way device with its own service life, accessibility requirement and code restrictions, and it cannot relieve positive pressure at all - the AAV article covers where it is and is not the right answer.

The case: a lavatory that gulps after a remodel

A powder room lavatory, no complaints for years, moved along the same wall during a remodel. Within a month, a gulp at the end of every use and an intermittent odor.

What was measured. Developed length from crown weir to the vent connection, following the fittings: 8 feet 2 inches, which is 8.17 feet. Slope checked with a level along the arm: a quarter inch per foot, as required. Arm diameter: 1-1/2 inches.

Fall against the allowance. 8.17 feet times 0.25 inches per foot is 2.04 inches of fall. The allowance is one diameter, 1.5 inches. The arm is at 2.04 divided by 1.5, which is 1.36 diameters, so it is 36 percent past the point where the crown weir stays above the vent invert.

The same finding as a length. 1.5 inches of diameter divided by 0.25 inches per foot gives a 6.0 foot maximum. The arm is 8.17 feet, an overrun of 2.17 feet. The adopted table for 1-1/2 inch pipe reads 6 feet, so the derivation and the table agree, and on the permit it is the table that governs.

The seal loss, measured rather than assumed. Trap refilled to its nominal 2.0 inches. The lavatory run alone, nothing else in the building: re-dipped at 0.7 inches. Loss 1.3 inches on the fixture's own discharge. Probe resolution is 0.06 inches, so that is a real reading by a wide margin, and because it happened on the fixture's own use with nothing else running, it is self-siphonage rather than induced.

What each candidate fix would have done to that number.

• Move the vent connection to 5.5 feet. Fall becomes 5.5 times 0.25, which is 1.375 inches, or 0.92 diameters. Inside the allowance with a little margin. This is the fix that was done.
• Upsize the arm to 2 inches and leave the length. Fall is unchanged at 2.04 inches and the allowance rises to 2.0 inches, giving 1.02 diameters. Still over, by four hundredths of an inch. When the entire allowance is one diameter, being slightly over is not a rounding question, and the 2 inch table limit of 8.0 feet says the same thing about an 8.17 foot arm.
• Flatten the slope. Holding the fall to 1.5 inches over 8.17 feet needs 0.18 inches per foot, below the required minimum for that size. Not available, and it would trade a siphonage problem for a solids problem.
• Fit a deeper trap. At 1.3 inches lost per event, a 3 inch seal would come back at 1.7 inches on the first use and be gone in three. It buys two extra uses, not a fix.

Cutting the new vent in. Support the pipe on both sides of every cut before cutting: on cast iron a snap cutter releases the chain's stored energy the moment the pipe parts and an unsupported run drops. On plastic, solvent cement and primer give off solvent vapor, so work with the space ventilated as the product's safety data sheet requires and never in a closed crawlspace without forced ventilation, and never apply a torch or heat gun to PVC or CPVC to soften or bend it, because heated PVC releases hydrogen chloride.

Confirmation. Trap refilled to 2.0 inches, lavatory run alone: re-dipped at 1.95 inches. The 0.05 inch change is under the probe's 0.06 inch resolution, so it is recorded as no measurable loss, not as zero. Three more uses over the next hour, same result.

The failure mode. The wrong call available at every step here is the one the customer suggests: the drain is slow, so snake it. The arm was clear the whole time. Snaking a clear arm produces a clean camera picture, a satisfied customer for a week, and a second visit, because the fault was a dimension rather than an obstruction.

How to verify the fix rather than the symptom

Measure the developed length again after the work, not before, and write it on the ticket with the slope you actually achieved. An arm cut in at 5.5 feet and then hung to 3/8 inch per foot by a helper is 2.06 inches of fall and back over the allowance, and nothing about the finished wall will tell you.

Then re-run the event test with the trap refilled and the fixture used alone, three times, reporting any change smaller than your probe's resolution as no measurable loss. One clean use is not a result: self-siphonage depends on how completely the arm fills, which depends on how fast the basin was drained, so test it the way the customer uses it, full basin, plug pulled.

Last, test the neighbours. Cutting a vent into a branch changes the air path for every trap on it. Refill and re-dip the other traps on that branch after the work and before you leave, because the visit that fixes one fixture and quietly breaks its neighbour is indistinguishable, from the customer's side, from a repair that failed.

References

• The plumbing code as adopted and amended by your local jurisdiction (the IPC or UPC family), for maximum trap arm developed length, minimum slope and vent connection rules; it binds through the adopting ordinance and the permit rather than on its own
• Safety data sheet for the solvent cement and primer in use, for the ventilation required when cementing in a confined or unventilated space
• See related: Drain Pipe Slope + Venting Quick Reference; Air Admittance Valve (AAV) vs Traditional Vent Stack Reference; How a Trap Seal Is Lost and How to Tell Which Way It Went; What Back Pressure Does at the Base of a Stack