There's a moment on almost every crew when the rope stops moving. One person is at the far box leaning back with their whole weight. Another is at the feed end guiding wire into the pipe. Somewhere between them, behind drywall or under a slab, the conductors have stopped. Nobody says it, but everyone is thinking the same thing: I can make this go. Pulling harder feels like effort, like refusing to quit. Usually it's the worst choice in the room, and the cause was settled weeks earlier, when someone bent one more ninety into the run than the physics could take.

The National Electrical Code has a rule for that moment. It reads like bureaucracy: no more than 360 degrees of bends between pull points. It's actually a short lesson in how friction behaves, and most of us learned the wrong version of it in school.

The rule, in plain language

For EMT, NEC 358.26 says there can't be more than the equivalent of four quarter bends, 360 degrees total, between pull points such as conduit bodies and boxes. The same limit shows up in the articles for rigid metal conduit, IMC, PVC, and most other raceways, each in its own .26 section.

Ask a room of electricians why the number is 360 and you'll mostly hear guesses: tradition, a round number, something a committee settled on. The real reason is more interesting, and once you understand it you'll lay out conduit differently.

Friction doesn't add up. It multiplies.

The physics you probably remember goes like this: friction equals the coefficient of friction times the normal force. Drag a box across the floor and the resistance stays steady. On a straight run of conduit, pulling tension behaves roughly that way. It grows with the weight of the wire and the length of the run, one foot at a time.

A bend works differently. When wire turns a corner under tension, that tension presses the wire into the inside wall of the bend. More tension means more pressure, more pressure means more friction, and more friction means more tension is needed to keep the wire moving. The effect feeds itself.

Engineers describe this with the capstan equation, the same math that explains how a sailor with a couple of wraps of rope around a winch drum can hold a load far heavier than their arms could. Tension leaving a curve equals the tension entering it multiplied by e raised to the coefficient of friction times the angle of wrap in radians. The variable that matters is the angle, and it sits in the exponent.

So each bend multiplies the tension instead of adding a fixed amount to it.

Here's an example. Take a coefficient of friction of 0.5, which is in the range for a poorly lubricated pull. Real values depend on the jacket, the raceway, and the lubricant, so treat this as a sketch of the curve, not a design number. A single 90-degree bend is about 1.57 radians, so it multiplies the incoming tension by roughly 2.2. Two quarter bends give about 4.8 times. Four give about 23 times. A fifth would push it past 50.

With good lubricant the coefficient drops and the numbers shrink a lot. The shape of the curve stays the same, though. Every additional bend costs more than the one before it. The fourth ninety isn't just one more ninety. It multiplies everything the first three already built up.

That's why the code draws a hard line. Past roughly 360 degrees, the tension needed to finish a pull climbs fast enough to stretch conductors, strip insulation, or snap the rope.

Where the damage actually happens

The obvious failure is a pull that won't move. The quieter failure is a pull that does move.

Where wire bears against the inside of a bend, it takes what cable engineers call sidewall pressure: the crushing load of tension pressed into a curve. Insulation can scuff, thin, or cold-flow at that spot. The wire comes out the far end looking fine, gets terminated, passes inspection, and carries load for years with a weak spot nobody can see, tucked inside a bend where nobody will look.

This is the uncomfortable part. A job that was forced through can look exactly like a job that was laid out well. The difference is in the pipe, and it shows up later as a ground fault no one can explain.

Why we keep pulling anyway

There's a human reason the rope stays tight past the point where it should go slack. Behavioral scientists call it escalation of commitment. Organizational psychologist Barry Staw studied it in the 1970s, finding that people who feel responsible for a course of action tend to put more resources into it after it starts failing, not fewer. The effort already spent starts to feel like a reason to keep spending.

You can feel this at the end of a stuck pull. Forty minutes of feeding wire, a crew watching, and an apprentice who doesn't want to be the one who says stop. Backing out means admitting the layout was wrong, and that's hard to admit. Pulling harder feels like strength, even when it's really just fear of losing face.

The 360-degree rule works partly because it takes that decision away from you in the moment. It moves the judgment to the layout stage, when you're calm and no rope is involved yet.

The bends you forgot to count

Most runs that go over 360 degrees don't fail because of four obvious nineties. They fail because of small bends nobody added up.

An offset around a beam is two bends. Two 30-degree bends make 60 degrees, even though the pipe ends up pointing the same direction it started. A three-point saddle over another pipe, with a 45 in the middle and two 22.5s on either side, adds 90 degrees, a full quarter bend, to a run that still looks straight. A kick to clear a box adds more. Add a couple of offsets and a saddle to three nineties and you're well past the limit on a run that looked reasonable from the ladder.

The code counts all of it, and so does the friction. The capstan equation doesn't care about the direction the pipe ends up pointing. It only cares how much total angle the wire had to wrap around.

Feed from the end nearest the bends

The multiplying effect has a practical consequence that runs against intuition. A bend multiplies whatever tension enters it, so where the bends sit along the run matters.

If the bends are near the end you're pulling from, the wire reaches them carrying all the friction it picked up along the straight run, and each bend multiplies that large number. If the bends are near the feed end, the wire passes through them early, while tension is still low, and the long straight stretch afterward adds tension one foot at a time instead of multiplying it.

It's the same pipe, the same wire, and the same bends, and the pull can be dramatically easier depending on which end you start from. Experienced pullers learn this by feel. The math explains why.

Your next moves

  • Total the bends before you bend anything. On your next run, list every planned bend between pull points, including offsets, saddles, and kicks, with its angle. Add them up. If you're over 360, add a pull point now, not after the wire is stuck.
  • Use 270 as your working limit. 360 is the legal maximum, not a target. On long runs, heavy conductors, or tight fill, put in a conduit body or pull box once you pass about three quarter bends' worth.
  • Choose the feed end on purpose. Look at where the bends are clustered and feed from the end closest to them. Tell the crew why so the habit spreads.
  • Use lubricant on any pull with more than one ninety. Lowering the friction coefficient shrinks every multiplication in the run. Use a lubricant that's compatible with the conductor insulation.
  • Agree on a stopping point before the pull. Settle in advance that if the rope stalls, everyone stops and backs off instead of adding muscle. Saying it before the pull makes it much easier to say during the pull.

Laying out the run before you bend it

The best time to solve a stuck pull is before any pipe is bent, when you're looking at the run and adding up angles. That's what the conduit-bending tools in Voltly are for: they work out offsets, saddles, and multipliers so you know exactly how many degrees you're putting into a run while you can still add a pull point. It works offline next to the ampacity, box fill, and voltage drop calculators, so it's there in a basement with no signal. If you'd like the math with you on your next layout, you can find it at voltly.lumenlabs.works.