Torque wrench

How it works

A wrench turns a fastener with the force the hand applies multiplied by the length of the handle, and a torque wrench measures that product. Sharp’s 1931 patent states the aim plainly: a wrench that indicates the amount of torque used in tightening or loosening a nut, with a pointer that moves over a scale as spring means are stressed.

Three principles appear in the patents and manuals. In a beam wrench, a steel beam between the head and handle bends under load, and a pointer fixed to the head reads against a dial on the handle, as in Sturtevant’s 1938 application, which says the beam is never loaded beyond one-half of its elastic limit so that it takes no permanent set. In a click wrench, a mechanism releases at a preset torque; the TEKTON manual describes a roller bearing and roller pin that roll over each other at the set value, producing a click that can be heard and felt. In the Bahr and Pfefferle wrench of 1935, a second handle member moves against a stiff spring, and the movement beyond a predetermined point is communicated to the operator.

The user sets or reads the torque, applies steady force at the grip, and stops. With a click wrench the setting is made on a scale on the handle; the TEKTON manual describes a coarse main scale with a finer micrometer scale on the knurled handle and a lock nut that fixes the setting.

Anatomy

Head with drive square: Takes the socket. Sturtevant’s 1938 application describes a head little larger than the diameter of an ordinary socket, so that the wrench can be used in limited space.
Measuring beam or torsion member: In a beam wrench, a steel beam that deflects under load; Sturtevant’s patent reduces its cross section outward from the head to spread the strain.
Pointer and dial: On a beam wrench the pointer is fixed to the head and reads against a dial fixed to the handle, which Sturtevant says shows tension in either direction.
Release mechanism: On a click wrench, a roller bearing and roller pin (per TEKTON) that roll over each other at the set torque.
Scale and lock nut: A main scale and a micrometer scale on the handle set the torque; a lock nut holds the setting.
Handle grip: The knurled handle is the point where the user is meant to hold; TEKTON says to grasp the knurled handle only and apply a pulling force.

Variants

Indicating (pointer-and-scale) wrench: The earliest patented form: Sharp’s 1931 application uses springs stressed by the load and a pointer moving over a scale.
Beam wrench: A deflecting steel beam with a pointer and dial; Sturtevant’s 1938 application describes one with no gears, coil springs or levers.
Click (preset or limiting) wrench: Releases with a click at a set value; TEKTON’s manual describes a dual-range model marked in foot-pounds and newton meters.
Ratchet-type signaling wrench: The Bahr and Pfefferle wrench of 1935, a ratchet socket wrench with a second handle member that moves against a spring and signals the predetermined torque.
Torque screwdriver and electronic tester: The Hand Tools Institute lists ASME B107.300, Hand Torque Tools and Torque Testers, among the B107 standards.

Specifications

Accuracy · percent of reading: +/- 4 (one maker’s click wrench)TEKTON states its model 24340 is calibrated and tested to +/- 4 percent. No standard tolerance has been read here.
Torque range · ft-lb; N·m: TEKTON model 24340 spans 25 ft-lb (33.9 N·m) upward; it does not measure below the lowest marked value. Ranges are model specific.
Drive size · in: The square drive follows the socket drive sizes; see the socket wrench entry.

Made to

ASME B107.300
ISO 6789

Safety and care

Safety

Do not use a torque wrench to break loose a stuck fastener; TEKTON’s manual prohibits it.
Do not increase the leverage of the wrench with another device; the TEKTON manual warns this can damage the wrench and cause injury.
Do not continue to pull after the click; the manual says doing so over-tightens the fastener and can damage the wrench.

Care

Store the wrench at its lowest torque setting (25 ft-lb on the TEKTON model) and do not turn the handle below the lowest setting; TEKTON says this relieves tension on the internal spring and prevents fatigue that affects accuracy.
Treat the wrench as a precision measuring instrument and store it in a clean, dry place (TEKTON).
Wipe it with a clean, dry, lint-free cloth and do not immerse it in liquid or cleaner (TEKTON).
If it has not been used for a long time, operate it several times at a low setting so the internal lubricant recoats the moving parts (TEKTON).

Tips

Hold the handle at the grip and pull
For accurate operation TEKTON says to grasp the knurled handle only and apply pulling force. The manual’s figure marks the handle pivot point and the click.1
Slow to a steady pull as the fastener gets snug
TEKTON advises operating the wrench like an ordinary socket wrench until the fastener becomes snug, then slowing to a smooth, steady pull, because working too fast or with too much force can overshoot the setting.1
Stop at the click and release the handle
At the instant the wrench clicks the torque setting has been reached. Releasing pressure resets the mechanism automatically for the next fastener.1
Be careful with the click at low settings
TEKTON warns that at low torque settings the click can be subtle, and recommends a quiet environment and extra care in releasing the handle at the right moment.1
Read the two scales together
The main scale gives the coarse value and the micrometer scale on the knurled handle divides it finely; in TEKTON’s example 100 ft-lb on the main scale plus 20 on the micrometer scale sets 120 ft-lb. Lock the setting with the lock nut by hand.1
It measures in the clockwise direction only
TEKTON’s manual marks the wrench rotation and states that torque is measured in the clockwise direction only, which is why it is used for tightening and not for loosening.1
Use the wrench inside its range
TEKTON’s model 24340 does not measure torque below its lowest marked setting of 25 ft-lb, so a lower specified torque calls for a wrench of lower range.1
A beam wrench reads in either direction and has nothing to reset
Sturtevant’s application says the dial of his beam wrench indicates tensions in either direction, and that it has no coil springs or gears; the beam is stressed less than half its elastic limit, so it does not take a permanent set.1

History

Earliest documented

18 May 1931 (documentary, filing)

Chicago, Illinois

John H. Sharp of Chicago filed US application 538,292 on 18 May 1931 for a “torque-measuring wrench” that indicates the amount of torque applied in tightening or loosening a nut by a pointer moving over a scale; the patent, 2,007,880, issued on 9 July 1935. It is the earliest torque wrench patent located in this research. A claim that Conrad Bahr built a torque limiting tool in 1918 is not supported by a patent or a document read here.1 2

The torque wrench was a response to uneven tightening. Bahr and Pfefferle’s 1935 application states the problem for bolted pipe joints: a circular series of bolts and nuts compresses a packing, the operator often works in a ditch, and unequal bolt tension can destroy the seal, strain the metal parts, strip threads or fracture bolts. It adds that operators differ in strength and in the leverage they apply, so that uniformity cannot be left to the individual. Sturtevant’s 1938 application makes the same case for engines: the best of mechanics, he writes, completely fail to equalize force by feel, and unequal head stud tension warps cylinder blocks.

The patent record shows several designs arriving within a few years. Sharp filed first, in 1931, with a spring-loaded pointer wrench that he also proposed for comparing different lock nuts under identical loads. Bahr and Pfefferle filed in 1935 for a ratchet wrench that signals a preset torque while the socket-carrying part stays connected to the handle; the patent says that if the operator moves the reversing lever to neutral, the failure of the pawls to slip on a backward movement tells him the predetermined torque has been reached. Sturtevant filed in 1938 for a beam wrench, and wrote that the tools then on the market were complicated and could be damaged. The three sources are patents; they do not tell who first built a torque wrench, and none mentions an earlier tool.

The popular account of the origin is not supported by the patents. A forum post, summarizing a retailer’s history page, says Bahr invented the torque wrench in 1918 and that Chrysler developed a beam type in about 1938 and licensed it to a small company. The documents read here show Bahr’s application dated 1935, a patent issued in 1937, Sharp’s earlier 1931 filing, and Sturtevant named as inventor of the beam wrench application of 1938, with no mention of Chrysler. A 1918 tool, if it existed, is not documented in the sources located here.

The change that mattered after the first patents was the click. The TEKTON manual shows how a modern preset wrench works: a roller bearing and roller pin roll over each other at the set value, and the user feels and hears the release. The same manual, in common with the Bahr patent, relies on the ear, which is why it warns that the click can be subtle at low settings.

O*NET’s Tools Used table lists torque wrenches in 6 of the 24 occupations mapped to this archive’s trades.

Timeline

1918
Popular accounts say Conrad Bahr of the New York City Water Department invented a torque limiting tool this year to equalize the tightening of flange bolts; no patent or contemporary document has been located (see disputed).1
1931
John H. Sharp of Chicago files an application for a torque-measuring wrench with spring-loaded arms, a pointer and a scale (18 May).1
1935
Sharp’s patent, US 2,007,880, issues on 9 July. On 12 June Conrad Charles Bahr of New York and George H. Pfefferle of Bradford, Pennsylvania, file an application for a ratchet-type torque measuring wrench, assigned to S. R. Dresser Manufacturing Company.1 2
1937
US Patent 2,074,079 to Bahr and Pfefferle issues on 16 March; it is for use on bolted pipe joints and couplings, where uneven bolt tension can ruin the seal.1
1938
Paul A. Sturtevant of Elmhurst, Illinois, files an application for a beam torque wrench with the pointer fixed to the head and a dial on the handle (28 March).1
1942
Sturtevant’s patent, US 2,283,707, issues on 19 May. In it he says that all measuring wrenches on the market depend on small gears, glass dials, actuating rods, pull strings or similar mechanisms.1
2021
ASME publishes B107.300-2021, Hand Torque Tools and Torque Testers, as listed by the Hand Tools Institute.1

Where the sources disagree

Who invented the torque wrench, and when?

Conrad Bahr invented the first torque limiting tool in 1918 while working for the New York City Water Department.1
Popular accounts as quoted on the Model T Ford Club of America forum
Bahr and Pfefferle applied for their patent on 12 June 1935 and it issued on 16 March 1937; the patent gives no date of invention.1
US Patent 2,074,079
John H. Sharp applied on 18 May 1931 and received US Patent 2,007,880 on 9 July 1935.1
US Patent 2,007,880

When and by whom was the beam torque wrench patented?

Chrysler developed the beam type around 1938 and licensed it to the small company that became Sturtevant Richmont; Sturtevant patented it in 1938.1
Popular accounts as quoted on the Model T Ford Club of America forum
Paul A. Sturtevant filed on 28 March 1938 and the patent issued on 19 May 1942; it does not mention Chrysler.1
US Patent 2,283,707

Gaps in the record (7)

No document supports the 1918 date for Bahr’s first tool; it appears only in popular accounts.
Sharp’s, Bahr’s and Sturtevant’s patents are the only primary sources; no earlier torque wrench patent has been searched for systematically because Google Patents is not available from this machine.
The origin of the click-type wrench has not been sourced; the inventor and patent are not identified.
Sturtevant Richmont’s own company history refused access from this machine.
ISO 6789 and ASME B107.300 have not been read; the accuracy class and calibration interval are not given.
Care and tips rest on one manufacturer’s manual and one patent for a particular design.
No independent test data has been found, so buying is null.