Torque Wrench

Torque = force × perpendicular distance from the axis. Newton-metres. Write it down before touching a tool.

Clamp a Grade 8.8 hex bolt upright in the bench vise with a nut run down finger-tight.

Fit the correct socket. Measure from the bolt axis to the point on the handle where you will pull, in metres.

Hook the spring scale at that marked point and pull perpendicular to the handle until the nut just turns. Read the force in newtons.

Multiply force by arm length. Example: 60 N at 0.25 m = 15 N·m. Record it.

Move your pull point to half the distance and tighten to the same 15 N·m. You now need twice the force. Confirm it on the scale.

Without any scale, tighten the nut to what feels like 15 N·m. Have someone measure each attempt with the spring scale. Record all five.

Largest guess minus smallest, as a percentage of the target. A spread of 30-50% is normal, and it is the entire case for the instrument.

Set the torque wrench to 15 N·m and tighten the same nut until it clicks. Note that the click is a release, not a stop.

Loosen, then re-tighten with the spring scale until the wrench would click. Compare the two numbers. Two independent methods, one quantity.

Put a drop of oil on the threads and tighten to the same 15 N·m. The nut turns visibly further — same torque, more clamping force, because friction fell.

Measure the bolt's length with the micrometer before and after tightening to 15 N·m. The difference — a few hundredths of a millimetre — is what actually holds the joint.

Try the same feel-test on a #6-32 machine screw with a hex key. Small fasteners are where feel fails worst, because the range between loose and stripped is tiny.

US 2,074,079, “Torque Measuring Wrench”, Conrad Charles Bahr, New York, and George H. Pfefferle, Bradford, McKean County, Pennsylvania. Granted 16 March 1937. The patent addresses ratchet-type socket wrenches for “the assembly of bolted pipe joints, couplings, clamps, sleeves, repair devices and other fittings”.

Three dates that sources merge. Bahr conceived the tool around 1918 as a municipal engineer in New York, to stop workmen cracking water-main fittings by over-tightening. The first torque-wrench patent filing on record is John H. Sharp of Chicago, 1931. Bahr and Pfefferle filed in 1935 and were granted in 1937. “Bahr patented the torque wrench in 1918” is a compression of all three into one, and it is false.

What a torque wrench actually controls, and does not. The engineer wants a clamping force, achieved by stretching the bolt elastically — step 12 measures it directly. What the wrench measures is torque, and most of that torque is spent overcoming friction under the head and in the threads, not stretching anything. Change the lubrication and the same torque gives a different clamp load, which is why step 11 matters and why critical joints are specified by angle-of-turn or by direct bolt-stretch measurement rather than by torque alone.

The click. The familiar clicker wrench uses a spring-loaded cam that slips at the set value — it signals, it does not prevent. Keep pulling after the click and you keep adding torque. The earlier beam type, with a pointer over a scale, cannot be set wrong because it has nothing to set; it is still the more honest instrument, and calibration labs like it for exactly that reason.

Why 1918 and not earlier. Hand-fitted machines had joints adjusted by a craftsman who knew each one. Mass-produced assemblies handed the same joint to any worker, in quantity, at speed. Once the skill left the hand, the number had to enter the tool — the same argument that produced every other measuring tool in the machine shop.