A bolt is a spring

Here's the question that separates torquing from guessing: when you tighten a bolt, what are you actually making?

A bolt is a spring

Not a hook, not a pin — a spring. Tightening the nut pulls the bolt through the threads and stretches it, a few thousandths of an inch. Steel stretched wants to pull back, and that pull squeezes the joint:

The joint holds because the plates are clamped — pressed together hard enough that friction between them carries the load. The bolt's threads don't "grip" anything; they're just the mechanism that loads the spring.

Torque is a stand-in

What you want is clamp force. What your wrench measures is torque. The two are related by friction:

• Most of your wrench effort — around 85–90% — is spent overcoming friction under the head and in the threads.
• Only the small remainder actually stretches the bolt.

That's why torque specs work at all (same bolt, same condition → same friction → same stretch) and why they're fragile (change the friction and the same torque gives a very different clamp).

Why joints fail

Two classic deaths, both about clamp, not threads:

• Under-clamped: the joint can slip and the bolt sees bending and shock it was never meant to carry — it fatigues and snaps, and everyone blames "a bad bolt."
• Over-clamped: the bolt is stretched past its yield; it never pulls back the same, and the clamp quietly dies.

A loose bolt almost never failed at the threads. It failed at the clamp.

Why this matters

Once you see clamp force as the real product, every torque-spec rule makes sense: why lube changes the number, why you never reuse torque-to-yield bolts, why the pattern and sequence matter on a flange. That's the next lesson.