Why a Tapered Thread Seals and a Straight One Does Not

Why this matters

Two threads can look identical in your hand, take the same wrench, start by hand the same way, and be designed to do opposite things. One is built to seal on its own threads by squeezing itself tight. The other is built specifically not to seal, because its job is to hold parts together while a separate element does the sealing.

Mixing them up produces a joint that assembles, holds a quick pressure test, and fails on the job. It also produces the more common and more expensive version: a tech who applies tapered-thread habits to a straight-thread fitting, or straight-thread habits to a tapered one, and cannot work out why the same technique leaks on one and splits the other.

What a straight thread refuses to do, on purpose

Start with the one people find strange, because understanding the refusal explains the taper.

A straight thread has the same diameter at the first engaged thread and at the last. Screw it in as far as you like and nothing gets tighter radially. There is clearance between crest and root the whole way, so there is a continuous spiral channel from one end of the engagement to the other, and no amount of turning closes it.

That is not a shortcoming. It is what makes the fitting good. Because the threads never generate radial interference:

• The threads carry load only. Tension, and that is all. No hoop stress in the female, which means no wall thickness spent on being squeezed and no split fittings.
• The joint is repeatable. It makes up to a position, or to a torque against a shoulder, and it makes up to that same position the next hundred times. Nothing about it depends on how many turns it went last time.
• The seal is a separate, replaceable element. An elastomeric ring in a groove, a bonded washer under the head, a gasket on a face. That element seals against pressure with a defined squeeze, and when it wears out you replace a part rather than a fitting.
• It comes apart and goes back the same. No progressive tightening, no accumulated interference, no limit on the number of remakes coming from the thread itself.

The straight thread splits the two jobs, mechanical holding and sealing, and gives each to whichever geometry is good at it. Every characteristic on that list follows from the refusal to seal on the threads.

What a taper is buying, and what it spends

A tapered thread changes diameter along its length at a defined rate. On the inch-based tapered pipe thread family, that rate is 1 in 16 on diameter: the diameter changes by one sixteenth of an inch for every inch of length, which is three quarters of an inch per foot. Both members are cut on that same taper, so as the male advances into the female, the male is being squeezed and the female is being stretched.

That interference is the seal and the mechanical joint at once. One geometry, both jobs. It is why a tapered joint needs no separate sealing element, why it works in a fitting with no room for a groove, and why the whole trade uses it.

What it spends is the strength of the female fitting, and it spends it in a way you can compute.

Axial travel per turn is one divided by the threads per inch. On a 14 threads-per-inch thread, one turn is 0.0714 in of travel. Over that distance the taper closes the diameter by 0.0714 divided by 16, which is 0.00446 in on diameter, or 0.00223 in of radial interference per turn. Those numbers are pure geometry of a 1-in-16 taper at 14 TPI and they do not vary with technique, material, or sealant.

Now put them on a fitting. Two turns past hand-tight is 0.00446 in of radial displacement. On a female thread whose pitch radius is around 0.375 in, dividing displacement by radius gives a nominal hoop strain around 1.2 percent. Common structural metals yield well below half a percent of strain. So the arithmetic is telling you something important: the joint cannot possibly accommodate that elastically, and it does not. The male compresses, the female stretches, and the thread flanks themselves deform plastically until the load redistributes. Tapered pipe threads seal by deforming, which is a design feature and also the reason for every limitation below.

What follows from the deformation

Make-up is one-way. Every remake into the same female drives further, because the plastic deformation from last time did not come back. A joint remade several times reaches a point where hand-tight engagement is deep, remaining travel is short, and the fitting is nearer its limit than a fresh one. There is a practical limit to remakes and it belongs to the fitting manufacturer, not to a rule of thumb.

Feel is a poor control. Torque is split between useful interference and friction, and a sealant that lubricates lets the joint travel further at the same wrench effort. A sibling article covers that in detail; the geometry here is why it matters, because extra travel converts one-to-one into extra interference.

The failure lands on the female. Hoop stress in a thin, brittle, or plastic female is what splits, and the split may not appear at make-up. A fitting stressed near its limit survives assembly and cracks on a pressure cycle or a cold morning, which reads in the field as a bad part.

Thread engagement, not thread count, is what holds. A tapered thread that bottoms out with no interference left, or one started crooked so only a few flanks touch, has neither seal nor strength however many turns it took.

Telling them apart with what is in your pouch

You do not need a gauge to answer the question in the field. You need a caliper and two measurements on the same thread.

Measure the major diameter at the end thread and again four threads back. On a 1-in-16 taper, four threads at 14 TPI is 4 divided by 14, or 0.2857 in of length, and the diameter should differ by 0.2857 divided by 16, about 0.018 in. A straight thread over the same span reads the same twice, within your measurement error.

The finding is the difference, not either absolute value, and that matters because absolute diameters get compared against the wrong nominal size constantly. If the two readings differ by roughly the predicted amount for the pitch you have, it is tapered. If they are the same, it is straight, and it needs a sealing element somewhere other than the threads.

Second check, on the female: look for a machined face, a chamfered counterbore, or a groove at the top of the thread. A straight-thread port that seals with a ring has a place for the ring to sit, and once you know to look for it you never miss it.

Worked example: the port that leaked with more tape every time

A hydraulic power unit with a pressure gauge port that had been leaking through three visits. Each visit added more tape and more wrench. Third visit the female threads in the manifold were visibly deformed at the top and the gauge fitting turned in noticeably further than it had the first time.

Before touching it: the unit was locked out and the accumulator and the circuit bled to zero under 29 CFR 1910.147, with the pressure proved at the gauge port itself and not at a gauge on the other side of a valve. A threaded fitting backed out of a pressurized port is a projectile, and a hydraulic stream can inject through skin without a cut.

Measurements, taken with a caliper on the removed fitting and on a plug from the same box:

The male is tapered: 0.018 in over four threads at 14 TPI is what a 1-in-16 taper predicts, 0.0179 in. The port, checked with a plug of known straight thread that entered full depth by hand with no increase in resistance and stopped against a machined counterbore with a groove in it, is a straight thread port designed for an elastomeric ring.

So the joint was a tapered male in a straight female. Only the last few threads of the taper were touching anything, the tape was carrying the pressure across a channel it was never going to close, and each visit's extra make-up worked the thin engaged section of the port harder. Three visits of correct technique applied to the wrong premise.

The fix was the correct straight-thread fitting with a new ring, torqued against its seat to the manufacturer's value, with no tape anywhere on it. The confirmation was the second-order detail: it made up to a defined position with the gauge face landing square, which is exactly what a straight thread is supposed to do and what none of the three previous attempts had produced.

The failure mode if you get this wrong the other way, straight male into a tapered female, is worse, because it feels right. It threads in a long way, gets snug, and holds a static test on the tape. There is no radial interference anywhere, so the only thing between the system and the room is sealant in a channel, and it lets go on a thermal or pressure cycle rather than on the bench.

How to verify you got this right

• Two caliper readings before the fitting goes in, whenever a joint has leaked more than once or the parts came from different sources. It takes fifteen seconds.
• A straight-thread joint made up against its seat with the sealing element present and undamaged, torqued to the manufacturer's value. If you found yourself reaching for tape on a straight thread, stop and find the element you are missing.
• On a tapered joint, count turns past a marked hand-tight and compare against the fitting manufacturer's range for that size and material, rather than judging by feel across materials that do not behave alike.
• If you smell gas at any point while working on fuel gas piping, everyone leaves immediately, no switches touched, no lights, no phone used inside, and the call goes out from outside the building to the gas utility. Leak test only with an approved method, never a flame.

References

• ASME B1.20.1, pipe threads general purpose, in the edition adopted by the plumbing or mechanical code your authority having jurisdiction enforces, which is how it reaches you rather than on its own
• 29 CFR 1910.147, control of hazardous energy, for isolation and stored energy before opening a pressurized threaded joint
• Fitting manufacturer documentation for permitted turns past hand-tight, remake limits, and straight-thread port torque values
• See related: Thread Sealants and What They Are Actually Doing; Why Threaded Connections Leak; Cross-Threading Prevention