A Bolt Is a Spring: Clamp Force, Stretch and Why Engine Builders Measure Instead of Torque

Engine builders will spend $200 on a set of rod bolts and then refuse to torque them. They put a dial gauge across each bolt and tighten until it has grown by a few thousandths of an inch. That habit is not superstition — it falls straight out of one equation, and once you see it, every torque spec on this site reads differently.

This is the long-form companion to our video A Bolt Is a Spring. The video moves fast; this page slows down, shows the graphs, checks the arithmetic, and explains the tool the builders are using — where it goes on the bolt, how it works, and what else you can do with it. If you just want your car’s torque spec, the lookup is here. If you want to understand what that number is actually doing, read on.

A bolt is not a clamp. It’s a spring.

Here is the part nobody tells you: if a bolt is tight, it is already stretched. Turning the nut does not “squeeze” anything. The threads act as an inclined plane that drags the bolt longer, exactly as if you were pulling on a very stiff coil spring. A stretched spring pulls back — and that pull, dragging the two halves of the joint together, is the clamp force. No stretch, no clamp. It is the whole ball game.

The spring picture is not a metaphor. It is literally Hooke’s law — the 1676 law of springs — applied to a steel rod. Force equals stiffness times stretch:

For a bolt, the stiffness is not a mystery number from a catalogue. It is built from three things you can measure with calipers and look up in one table:

Four letters. Each one changes something concrete in your garage, so let’s take them one at a time — but first, a feel for the size of the numbers.

How much stretch are we talking about?

Take an M12 × 1.5 wheel stud, torqued to a typical 100 ft-lb. Its tensile stress area is 88.1 mm² (0.137 in²), steel’s modulus is 30,000,000 psi, and the clamped stack — hub flange, rotor hat, wheel — is about 40 mm (1.57 in) thick. Plug those in and the stud’s stiffness comes out at 2.62 million pounds per inch. At 100 ft-lb a dry stud like that stretches roughly 0.005 in — five thousandths, the thickness of one sheet of printer paper — and that paper-thin stretch is holding 13,089 lb of clamp on your wheel.

Two things to notice on that graph. First, it is a straight line: double the stretch and you double the clamp, all the way up to yield. Second, look where the operating point sits. 13,089 lb over 0.137 in² is about 96,000 psi of stress in the stud, and a class 10.9 stud yields around 136,000 psi. That puts the wheel stud at 70% of yield — which is exactly the preload target fastener engineers aim for on a reusable bolt. The torque spec was never a random number. It is this equation, pre-solved.

Term by term: what A, E, L and ΔL each do

A — cross-section area

A thicker bolt is a stiffer spring. No surprise. The only subtlety is which area: the threads cut into the shank, so engineers use the tensile stress area — roughly the area of a circle drawn halfway between the pitch diameter and the minor diameter. That is why an M12 × 1.5 (fine thread, 88.1 mm²) is a stiffer, stronger spring than an M12 × 1.75 coarse thread (84.3 mm²) of the same nominal size.

E — the stiffness of the steel itself

This is the surprise. Young’s modulus is a property of the material, not of the part, and for every carbon and alloy steel used in bolts it is the same 30 million psi (207 GPa), give or take a percent. A 30-cent Grade 2 bolt and a $20 ARP bolt stretch exactly the same amount under the same load. What the expensive bolt buys you is not stiffness — it is how far up the line it can ride before it yields.

Strength and stiffness are different things, and most of the internet has them confused. A Grade 8 bolt is not “tighter” or “harder to stretch” than a Grade 5 — it simply survives more stretch before it goes plastic. (Our bolt grades guide covers how to read the head markings.) The practical consequence: upgrading grade never changes how much a bolt stretches at a given torque. It only changes how much torque you are allowed to use.

L — grip length

A longer bolt is a softer, more forgiving spring — the stiffness is divided by L. File that away; it pays off in a minute, because it is the single biggest reason long bolts survive vibration and short bolts loosen.

ΔL — the stretch

The only part you actually want. Everything else on the page — torque wrenches, angle gauges, K-factors, stretch gauges — is just a way of trying to hit the right ΔL.

The math, in four lines

If you have ever wondered where the A·E/L comes from, here is the derivation. It is short.

1. Hooke’s law for a spring: F = k · x. Force is stiffness times displacement.
2. Stress is load spread over the cross-section: σ = F / A. Strain is stretch as a fraction of length: ε = ΔL / L.
3. Hooke’s law for a material says stress is proportional to strain, and the constant is Young’s modulus: σ = E · ε.
4. Substitute the definitions into line 3: F / A = E · ΔL / L. Solve for F: F = (A · E / L) · ΔL. Compare with line 1 and the bolt’s spring constant falls out: k = A·E/L.

Now the M12 stud, with real numbers, so you can check us:

So why do we torque bolts at all?

Because torque is the cheap way in. Nobody is putting a stretch gauge on a lug nut in a tyre shop. But look at what a torque wrench is actually measuring: the twisting effort at the handle, which has to overcome friction under the nut face and friction in the threads before any of it is left over to stretch the bolt. As our Shorts have been saying: roughly 90% of your torque dies as friction, and only the leftover becomes clamp.

That is where the nut factor K in the familiar T = K·F·d comes from: it is a single number standing in for both friction terms. Plain dry steel runs about 0.19–0.20; a light oil drops it to roughly 0.12; anti-seize to 0.10 or lower. Same wrench, same click, and the lubricated bolt ends up with nearly twice the clamp — which is how people snap “correctly torqued” studs. Torque is a guess about friction. Stretch is a measurement of the spring.

Read that ladder as a story about friction. A torque wrench trusts K completely, so it inherits all of K’s scatter. Torque-plus-angle only uses torque to seat the joint, then sets the stretch geometrically through the thread pitch (Figure 7 below), so most of the friction error drops out. Ultrasonic tools time a sound pulse down the bolt and back to measure its length change. And a stretch gauge simply measures the bolt with a dial indicator — friction cannot lie to a length.

The stretch gauge: where it goes, how it works, how to use it

This is the tool from the video’s opening line, and it is far less exotic than it sounds. A rod-bolt stretch gauge is a rigid C-shaped frame with a ball-tipped anvil at one end and a spring-loaded dial indicator at the other. You hook it over a bolt so the two balls sit on the bolt’s two ends, and the dial reads how far apart they are. Tighten the bolt, and the dial reads how much the bolt grew.

Where it sits on the bolt

Performance rod bolts — ARP, Carrillo, Manley, most forged-rod kits — come with a small conical dimple machined into the centre of the bolt head and another into the tip of the threaded end. Those are not decoration: they are seats for the gauge’s ball tips, so the gauge lands in the same spot every time and cannot skate off a rounded edge. The fixed anvil goes in the head dimple; the indicator’s plunger goes in the tip dimple. The rod cap has to be installed and the bolt threaded in for the tip to be reachable, which is why this works on rod bolts (the tip pokes through the cap) and not on a head bolt buried in a block.

How a dial indicator works, in one paragraph

Inside the round face is a plunger riding on a rack. The rack turns a tiny pinion gear, which through one or two more gears turns the needle. The gearing multiplies the motion: on a common 0.001 in graduation indicator, one full sweep of the needle is 0.100 in of plunger travel, so 0.005 in of bolt stretch is five clearly separated tick marks. Rod-bolt gauges usually carry a finer 0.0005 in indicator, so the same stretch is ten ticks and you can see a quarter-thousandth. The bezel rotates so you can spin the zero to wherever the needle happens to be sitting — that is the whole zeroing procedure. A spring keeps the plunger pressed lightly against whatever it touches, which is why the reading follows the bolt tip as it moves away.

Step by step on a set of rod bolts

1. Lube the threads and the underside of the bolt head with the fastener maker’s assembly lube (ARP Ultra-Torque, or whatever the spec sheet names). The stretch method does not need a known friction, but a consistent one keeps the torque you end up applying within the bolt’s comfort zone.
2. Install the cap and run the bolts in until they are snug — finger-tight plus a light nip with a wrench, just enough to seat the cap. Do not load them yet.
3. Hook the gauge over one bolt, ball tips in the dimples. Rock it gently to make sure it is fully seated and the needle settles to a repeatable reading. Rotate the bezel to zero the needle. Some builders note the free-length reading on a card instead of zeroing, so they can re-check the bolt after a run — both work.
4. Tighten the bolt in small steps with an ordinary breaker bar or wrench, re-seating the gauge and reading between steps. Stop when the dial shows the spec stretch. For most ARP rod bolts the sheet calls for around 0.0055–0.0060 in; some specify 0.0050–0.0055 or 0.0060–0.0065. The sheet in the box overrides anything on this page.
5. If you overshoot by a little, loosen and re-tighten; the bolt is still elastic. If it is more than about 0.001 in past spec, or the reading will not come back to zero when loosened, the bolt has yielded — replace it. A permanently grown rod bolt is a rod bolt that will keep growing.
6. Repeat for the other bolt on the cap, then go back and re-check the first one: tightening its neighbour can change the cap’s seating and steal a few ten-thousandths.
7. Write the torque you ended up needing on your build sheet. If the next bolt hits spec stretch at a wildly different torque, something — a burr, a dry thread, a bent bolt — is wrong.

Where you would reach for a gauge instead of a torque wrench

The honest answer is: anywhere both ends of the bolt are reachable and the clamp really matters.

• Connecting-rod bolts — the classic case. A rod bolt is the most highly loaded fastener in the engine, it sees reversing tension every revolution, and a cap that walks a thousandth spins a bearing. This is what the tool was invented for.
• Any bolt whose spec is given as a stretch — some aftermarket main studs and flywheel/flexplate bolts ship with a stretch figure. If the tip is exposed after assembly, use it.
• Re-checking a bolt after a run — if you recorded the free length, you can put the gauge back on a loosened bolt and see whether it has permanently grown. A torque wrench cannot tell you that; a length can.
• Diagnosing a friction problem — if you hit the torque spec but the stretch is short, the torque went into friction (dry thread, damaged thread, the wrong lube). If the stretch is over, the threads were slicker than the spec assumed. Either way the gauge told you something the wrench could not.

And the same 0.001 in dial indicator, on a magnetic base instead of the stretch frame, is the most-used measuring tool in engine building for jobs that have nothing to do with bolts:

• Crankshaft end-play — indicator on the snout, pry the crank fore and aft.
• Deck height and piston-to-deck — bridge across the bore, zero on the deck, read the piston at TDC.
• Camshaft lift and lobe centreline — on a lifter or pushrod with a degree wheel.
• Runout of a flywheel, flexplate, brake rotor, wheel hub or crank snout — anywhere a wobble matters.
• Ring-gear backlash in a differential, and finding true TDC with a piston stop.

A stretch gauge is a specialist frame around a general-purpose instrument. If you build engines, buy the gauge; if you only ever do a rod set once, borrow one and put the money into a good indicator and base instead.

Recommended tools

Some links below are Amazon affiliate links — as an Amazon Associate we may earn a small commission at no extra cost to you.

Four things the equation changes in your garage

1. Rod bolts: read the stretch, not the torque

Covered above. The gauge reads ΔL directly, friction drops out of the problem, and the ARP stretch figure on the sheet is the only number that matters.

2. Head bolts: torque-to-yield is the stretch method for the rest of us

You cannot get a gauge onto a head bolt, so manufacturers do the next best thing: snug the bolt to a small torque to seat the joint, then turn it a fixed angle. The angle sets a displacement through the thread pitch, and displacement is what we wanted all along. It also deliberately stretches the bolt past yield — which is why TTY bolts are one-time-use and why the LS head-bolt sequence reads 22 ft-lb, then +90°, then +90° again.

Notice those advances are three to four times the 0.005 in we calculated for the wheel stud. That extra travel is not all bolt stretch — a head gasket and a long aluminium head squash under load too — but it is why TTY bolts end up well past yield, and why the second-use warnings are not marketing.

3. The 50-mile re-torque

Freshly machined surfaces are not flat at the microscopic level. Under load the high spots crush and the joint “settles” — by a commonly quoted 0.0004 in or so for a typical steel-on-steel stack. Off a 0.005 in stretch, that is 8% of your clamp, gone, without anyone touching the bolt. Re-torquing after the first heat cycle or the first drive simply gives the spring its stretch back. This is why new wheels get a torque re-check at the interval the vehicle or wheel maker specifies, and why head gaskets that require a re-torque say so.

4. Remember L? Long bolts live longer

That same 0.0004 in of settling is a fixed distance. How much clamp it costs depends on how much stretch the bolt had to begin with — and that is set by L.

A half-inch-grip bolt hitting 13,000 lb only stretches 0.0016 in, so 0.0004 in of settling is a quarter of its preload. A 3 in bolt at the same load stretches 0.0096 in and loses 4%. On a vibrating joint that difference is the difference between a bolt that stays put and one that works loose every few hundred miles. It is why engine designers use long bolts with reduced shanks in critical spots, and why a spacer under a short bolt sometimes fixes a loosening problem that Loctite could not.

Putting it back together

• A — buy the right diameter and thread. Area sets the stiffness and the strength ceiling together.
• E — every steel bolt is equally stiff. Grades buy yield strength, not stiffness.
• L — longer is softer, and softer is more forgiving of settling and vibration.
• ΔL — the stretch is the whole point. Torque, angle, ultrasonic and the dial gauge are just four ways of trying to hit it, in ascending order of accuracy and cost.

And your torque spec? It is this equation, pre-solved for your exact fastener and the friction the engineer expected. Which is why the spec, the thread condition, and a real torque wrench used properly all matter — and why, when the bolt matters enough, the builders stop guessing and measure the spring.

Frequently asked questions

Keep going

Bottom line: A bolt is a spring, clamp force is stiffness times stretch, and every steel is equally stiff. Torque is a cheap estimate of stretch made through friction; angle is a better one; a dial gauge on the bolt is the measurement itself. Use the torque spec for everything that does not matter enough to measure — and measure the ones that do.