The cable behind that switch is folded like a paper clip. Someone needed it to fit in a shallow box, so they pushed the slack in, pressed the device on top of it, and tightened the screws. The wall plate went on flat. The light came on. The inspector, if there was one, saw a neat, flush switch.

Behind it, the cable has a crease you could cut yourself on. Nothing has failed yet. But the cable has been changed in a way that nothing at the surface will ever announce, and it will carry that change for as long as the house stands.

The code has a short rule about this, and the rule is really about geometry.

What the code actually says

NEC 334.24 covers nonmetallic-sheathed cable (Types NM, NMC, and NMS), the stuff almost everyone calls Romex. It says bends must be made so the cable isn't damaged, and that the radius of the inner edge of the curve can't be less than five times the diameter of the cable.

Take a common 14/2 or 12/2 NM-B cable. It is a flat oval, roughly a half inch across its wide side and a bit under a quarter inch on its narrow side. Five times the diameter puts the inside of the curve at about two and a half inches. That is a sweeping, lazy loop. It is not a fold.

The phrase to notice is inner edge. The rule measures the tightest part of the curve, the part under the most stress, and not the cable's centerline. It also leaves no room to bend the cable "just this once" behind a box.

The geometry of a bend

Pick up a piece of foam pipe insulation or a thick rubber hose and bend it. The outside of the curve stretches. The inside compresses. Somewhere in the middle there is a line that does neither, which engineers call the neutral axis.

How much each part stretches depends on a simple ratio: the distance from that neutral axis divided by the radius of the bend. Strain is roughly distance from the middle, over bend radius. This is a standard result from beam bending, not anything specific to wire.

That ratio explains the whole rule. Make the radius small and the same distance from the middle produces far more strain. As a rough illustration, take a cable about half an inch thick. Bent around a radius of two and a half inches, the outer surface of the jacket stretches somewhere under ten percent. Bend the same cable around a radius roughly equal to its own thickness and the outer surface is stretching on the order of a third of its length. These are back-of-envelope figures from the geometry, not test data, but the direction is unmistakable. Halve the radius and you roughly double the strain.

Plastic jackets and insulation tolerate gentle strain for decades. They do not tolerate large strain well. They whiten, thin, crack, or craze, especially when the plastic is cold or old.

What happens to the copper

Here is the part that surprises people. Annealed copper is soft, and it yields at very small strain, a fraction of one percent. So almost any bend permanently deforms the metal. That is not the problem. Electricians bend copper constantly and it forms fine.

The problem is that bending a wire back and forth hardens it. Copper work-hardens: as you deform it, the crystal structure fills with tangled defects and the metal gets stiffer and less able to stretch further. Fold a conductor sharply, straighten it, fold it again, and the metal at the crease can crack. You can feel this with a paper clip. The first bend is easy, and the third one snaps it.

A single tight fold behind a box may not crack anything. But it concentrates deformation in one spot, and that spot is also where the wire may be nicked, pinched by a device, or flexed again the next time someone swaps a switch. A cracked strand raises resistance at that point, which means more heat in a place with no airflow. That's a familiar story if you've read about nicked copper wire and loose connections: small local damage, concentrated heat, slow failure.

The insulation is the real casualty

In NM cable, each conductor has its own thermoplastic insulation, and the whole bundle sits inside an outer jacket with paper filler. When the cable is folded tightly:

  • The outer jacket on the outside of the curve thins and can crack.
  • The conductors inside have to shift, because they cannot all sit on the neutral axis. The ones on the outer side get stretched against their insulation, and the ones on the inner side get squeezed against the others.
  • The inside of the fold can pinch the insulation of one conductor against the sharp edge of another, or against a box, a clamp, or a device.

The result is thinned insulation with no outward sign. Voltage is not what this fails against on day one. It fails when moisture, heat, a vibration, or a nail from a later renovation finds the weak point. Insulation that was designed with a certain thickness and now has less of it is simply a smaller safety margin that nobody recorded.

Where this actually goes wrong

The fold shows up in predictable places.

Shallow boxes. Old-work boxes and shallow remodel boxes leave very little room. The temptation is to stuff slack behind the device and crush it flat.

Tight corners at studs. A cable run around a stud edge or turned sharply to get through a bored hole at an awkward angle.

Re-bending. A cable that has been bent to a position, pulled back, and bent again. Each cycle adds work hardening.

Staple and clamp points. Pulling a cable taut through a staple or clamp right at a bend. The clamp then holds the bend in a tight radius permanently.

The common thread is that all of these are done to make something fit, and fit is exactly what the rule limits.

Your next moves

  • Measure it. Next time you open a box, hold a tape or a coffee can against the cable's curve. A can about three inches across is a decent stand-in for the inner radius NEC 334.24 expects on common cable. If the cable is tighter than that, rework it.
  • Make the curve before the box, not inside it. Leave the sheath on well into the box, pre-shape a wide loop of the cable a few inches back, and then ease the slack in. Don't push a straight cable at a wall and let it fold where it wants.
  • Use a deeper box. If the slack won't fit without folding, go to a deeper box or an extension ring. Check the volume with a box fill calculation while you're at it, since a cramped box is often both a bend problem and a fill problem.
  • Never re-bend to fix a kink. If a section has a visible crease, whitened jacket, or a flat spot, cut it out and splice in a box rather than straightening it and hoping.
  • Staple away from bends. Keep staples and clamps on straight sections, and let the curve be free to relax into a gentle shape.

The rule behind the rule

The bend radius is a good example of a code requirement that looks fussy until you do the arithmetic. The five-times figure is not a magic number. It is a point on a curve where strain stays low enough that the insulation, the jacket, and the copper can all be treated as undamaged for the life of the building. Past that, the cable has simply used up some of its margin.

The same logic explains why conduit benders have a defined radius for each size, and why the cable types that are armored or shielded have their own, usually larger, multiples in the code. Whenever you force a material around a corner, the question is always how tight, and the answer is always a ratio.

Where a field reference helps

The ratios are easy to forget on a ladder with a cable in each hand, and the numbers for bending conduit, filling boxes, and sizing conductors live in different tables. That's the reason we built Voltly, an offline electrician's calculator and NEC reference that keeps bending, box fill, and conductor sizing a tap away even in a basement with no signal. If you'd like to see it, it's at voltly.lumenlabs.works. Either way, the next time a cable doesn't quite fit, give it the room it needs instead of the crease it will keep.