Torque Wrench

identity

• Name: Torque Wrench (also: torque-measuring wrench; click wrench / clicker; beam wrench; dial wrench, by mechanism). Umbrella terms: torque instrument, torque tool family.
• Family / category: Wrenches/Spanners family, the metering branch. Where the Adjustable Wrench and Impact Wrench give no reading of the turning force applied, a torque wrench is a hand torque instrument that indicates, signals or, in some designs, limits the torque applied to a fastener.
• One-line: A hand tool that helps an operator follow a specified joint procedure by showing or signalling a torque value, so "tight" becomes a checked number instead of a guess. It does not measure clamp load directly, and most click designs do not stop the fastener turning if the operator keeps pulling after the release.

The family, split by what the tool does (a rule for one class is not a rule for another):

The ISO 6789-1 taxonomy separates Type I (indicating) and Type II (setting) tools [manual, relayed] [TW50]. A ratcheting square drive is a convenience feature, not the torque-sensing mechanism. Manual gear multipliers, pneumatic/electric torque tools and hydraulic torque wrenches are related industrial systems treated separately below (see variations).

✦ how NOT to use it

not-for: routine woodworking joinery/assembly with no published spec; high-volume repetitive driving; freeing seized hardware.

dangerous: physical-injury hazards, by cause. The earlier version of this entry said torque-wrench work has low physical-injury risk and lacks socket and slip hazards. That is wrong and is withdrawn. Torque wrenches often use sockets and extensions under substantial load.

• Socket and bit failure or slip. A wrong, worn, incompletely seated or non-impact-rated socket can slip or fail under load. Controls: select the correct size and type, inspect before use, seat it fully, use impact-rated components where a powered-tool procedure requires them, support long extensions rather than letting them sag or flex, and wear eye protection where appropriate. HYTORC's safety guidance calls for correct impact sockets, inspection, trained operators and hands kept clear of reaction points [manual, relayed] [TW47].
• Sudden release and loss of balance. A fastener or tool that lets go suddenly can throw the operator off balance or drive a hand into adjacent work. Controls: a stable stance, slow controlled force at the designated hand position, a clear path for the hand if the tool releases, and stopping at the indication.
• Stored mechanical force unwinding the tool. IMCA documented a near miss in which a diver was asked to operate an ROV override on a spring-actuated subsea isolation valve with a hand-held torque wrench; a third-party technician stopped it. The hazard was sudden, uncontrolled unwinding of the valve's spring mechanism, which could have spun the wrench into the diver. IMCA's cited contributors included ambiguous drawing warnings that hand-held tools should not be used and an operations manual that never reached the engineers. It recommends not operating such an override with a hand-held torque tool unless the client or valve manufacturer clearly confirms it is safe [TW46]. This is specific to stored-energy equipment, scoped to what the source supports.
• Reaction-arm crush (manual multipliers). IMCA Safety Flash SF 13/23 (May 2023): a project engineer, present to verify torque settings, stepped in to help while a deck engineer torqued Grayloc clamp bolts using a manual torque wrench with a force multiplier. His left hand, resting on the multiplier's reaction bar, slipped and was pinched between the bar and the gooseneck, causing a hand injury. IMCA's root causes: supervisors did not intervene when an outside person joined the work, his briefing did not explain his role, he missed the deck engineer's toolbox talk, and he was neither trained nor qualified for the activity [TW33]. Norbar's handbook states the same hazard: "there is a risk of crushing between the reaction bar and work piece," and "keep hands away from reaction bar" [TW45].
• Powered and hydraulic systems. These add crush, projectile, hose and reaction-point hazards beyond a hand wrench's. They are for trained personnel following that equipment's procedure: inspect the wrench, drive, sockets, reaction components, hoses and couplings before use; check the reaction geometry; keep hands out of the reaction zone and clear of likely travel. The specific hose, coupling and pressure details [manual, relayed] rest on the reviewer's description of the Enerpac material [TW49][TW54].
• Before every use, generally. Confirm units, direction, target value, certified range, fastener condition, lubrication state and tightening sequence before loading the tool. After a drop, overload, improvised extension or suspected damage, take the tool out of critical service until it is checked under the applicable procedure.

damaging: the entry's most tool-specific misuse findings.

• Over-torque after the release. On a click wrench, continuing to pull after the release can over-torque the fastener, possibly without the user noticing [TW43][TW18].
• Storage. Leaving a click-type wrench wound up in storage can fatigue the spring and drift calibration; Park Tool's fix for its click wrenches is winding down to the lowest scale value, not zero [TW18]. Other makes and mechanisms differ; follow the maker.
• Wrong friction condition. Torque values are only valid for the specified friction condition; adding lubricant or anti-seize that the procedure does not permit changes the clamp load delivered at the same reading. The earlier "30 to 65%" figure came from a consumer source and is removed; no universal percentage exists.
• Outside the certified range. A tool used outside the range printed in its instructions or certificate may not be accurate and, on a click tool, may give a faint or absent release. The earlier "below 20% of range" rule is withdrawn as a universal; accuracy range is model- and certificate-specific.
• Unit confusion (ft-lb vs in-lb vs N·m), with a reported damaging case: a wheel-stud spec listed at an implausible 290.7 ft-lb broke the studs when followed literally [TW36][TW37] (forum and trade sourcing; treat as an illustration).

beginner-trap: following a universal order instead of the joint's own. Tightening order and number of passes are joint-specific. Cylinder-head bolting is the best-documented case of an out-of-sequence problem: tightening out of order bows or tilts the head and crushes the gasket unevenly [TW38]. The lesson is to follow the procedure for the joint, not to apply a star order or a fixed half-then-full staging everywhere; a procedure may require three or more passes, a numbered sequence, angle steps or no re-torque. The same applies to wheels and to multi-screw cutterhead inserts, where the cutterhead manual decides. A second trap: stopping at the sound of the click instead of the head's deflection [TW18].

lineage-misuse: using a torque wrench as a plain wrench or breaker bar. CDI's factory repair manual calls this the most common form of abuse among wrenches sent for repair, identifiable by a bent tube, and says a wrench showing this damage should be replaced, not repaired [TW15]. A cross-reference to the Impact Wrench entry: trusting an impact tool's feel instead of finishing a spec'd, safety-critical fastener with a calibrated check, of which Jesse Busenitz's account of shearing a jointer-insert screw with an impact driver is a woodworking-specific instance [TW19][TW39].

correct use

Scoped to the ordinary hand torque wrench or torque driver (indicating, setting or limiting). Multipliers, powered and hydraulic systems are trained-use equipment with their own procedures and are not covered by these steps.

• Get the procedure first. The target value, units, fastener condition, lubrication state, fastener sequence, number of passes, angle steps and re-torque rule all come from the fastener or equipment procedure for that exact joint. Where this entry and the maker's manual differ, the manual wins.
• Check units and range. Set the value in the correct units (ft-lb, in-lb, N·m; confusing them is a real failure mode, see below). Use a calibrated tool whose certified operating range, as printed in its instructions or calibration certificate, contains the target. If the target lies outside it, select a smaller or different tool.
• Seat the drive. Inspect the wrench, drive, socket or bit and any extension, and seat the correct size and type fully before loading.
• Hold and load as the maker directs. Apply force slowly and in a controlled way at the marked hand position; Park Tool says to centre the hand on the handle of a click wrench, and to pull where practical. Keep a stable stance and a clear path in case the tool or fastener releases.
• Stop at the indication. On a click wrench the deflection of the head, not the sound, is the signal, and Park notes a click may be faint or absent at low settings, so use ears and eyes; turning past deflection over-tightens the fastener [TW18]. Stop immediately at the indication or release unless the exact tool is designed to disengage or cam over.
• Sequence. Follow the exact fastener sequence, pass schedule, lubrication state, angle step and re-torque instruction supplied for that joint. Use a cross or spiral sequence only when that procedure calls for it. Sequences differ: cylinder heads, wheels, flanges, cutterhead inserts and machine guards are not interchangeable, some procedures call for three or more passes or numbered orders, some forbid re-torque, and some small inserts are seated one at a time. A forum post is a lead to a manual, not the operating authority.
• Friction condition. A torque value is valid only for the specified friction condition (clean and dry, oiled, plated, coated or lubricated). Never add lubricant or anti-seize unless the procedure permits it. Only a small share of input torque stretches the bolt (Enerpac estimates 10 to 15%, with 85 to 90% consumed by friction), which is why thread and coating condition matter [TW48].
• Witness marking, afterwards. A torque-seal or witness-paint mark applied after the specified tightening operation reveals later relative movement or tampering. It does not measure the original torque, retained preload or joint integrity, and it does not substitute for torque verification or inspection [TW16][TW17].
• Storage and verification. Follow the maker's instructions. For Park Tool's click wrenches that means winding to the lowest value on the scale, never to zero; for its limiter type no particular setting is needed; for a beam wrench Park gives none [TW18]. For other adjustable mechanical click models follow the manufacturer's storage setting and never wind below the marked minimum.
• Calibration. Verification frequency follows the manufacturer's instructions, the tool's certificate, frequency and severity of use, overload or impact events and the user's quality system. As a labelled example only, Park Tool recommends calibrating its click and limiter types every 5,000 clicks or at least annually for frequent use [TW18]. This is a Park Tool recommendation, not an ISO 6789 requirement [manual, relayed] [TW50]. Calibrate after a drop or suspected overload.

Deep operational sequencing belongs to Reed; this is the tool-level discipline.

✦ matching

does: a hand torque instrument that indicates, signals or, in some designs, limits the torque applied to a fastener, helping an operator follow a specified joint procedure. It does not measure clamp load and most click designs do not physically prevent over-torque [TW43].

suits: specified fasteners around a shop, recurring in cutterheads, machine assembly, vehicle and trailer work and other fasteners that have a manufacturer's torque specification. The best-documented woodworking case is jointer/planer knife or spiral-cutterhead carbide-insert screws. The consequence of over-tightening can include stripped inserts, damaged screws, distorted seats or fractured carbide, so this is a good Hewn example, but only conditional on the exact machine or cutterhead manual: the value, sequence, lubricant state, tool and replacement rule come from that model-specific manual (for example the Byrd, Grizzly or other maker's manual for that model), and no number is given here. Such work usually needs a small, low-range torque wrench or torque driver whose certified range contains the target. Other cases: torque-plus-angle procedures with an angle-capable tool; adjacent trades a woodworker might cross into, such as bicycle maintenance, small-engine and chainsaw work and automotive repair. Explicitly does NOT suit ordinary woodworking joinery or shop fastening with no published spec. For arbor nuts and similar machine fasteners, do not guess: consult the model-specific manual. An earlier version of this entry cited a search-synthesis line about SawStop's arbor nut ("snug, not spec'd"); it was not an independently fetched statement and is withdrawn.

scope bands (they do not inherit one another's access or safety rating): ordinary hand torque wrench; precision torque driver; manual multiplier; powered or hydraulic bolting system.

kit-fit:

• Pairs with: a drive-size-matched, correctly rated socket set (torque wrenches are sold bare-drive); a torque angle gauge for torque-plus-angle procedures [TW8]; a torque wrench tester, a distinct instrument used to verify the wrench, not to fasten anything [TW31].
• Substitutes for: nothing cleanly on a spec'd fastener where the procedure calls for a torque tool. Electronically torque-controlled "smart" impacts and torque-control power tools can be appropriate under a validated procedure, so the blanket claim that all critical work is finished by hand is not made here [TW32].
• Cannot replace: a breaker bar for freeing a seized fastener. Never use a torque wrench as a breaker bar unless its manufacturer explicitly permits it and specifies the limits; Park Tool also advises against using click-type wrenches as common ratchet handles to remove fasteners [TW18]. Nor a plain ratchet for routine, high-volume, non-spec'd fastening.

not-for (project-keyed): everyday joinery/assembly fastening with no published spec; freeing stuck, rusted or seized hardware; fast, repetitive driving.

access (tiered, replacing the earlier "low physical-injury risk" rating):

• Small manual tools (torque driver, ordinary hand click or beam wrench on small fasteners): generally accessible with instruction. Not risk-free: sockets, extensions and bits can slip, strip or snap, and a sudden release can throw the user off balance or drive a hand into adjacent work.
• High-leverage hand use (long handle, extension, high target torque): elevated. More stored energy, more reaction force and a worse outcome from a slipped socket.
• Reaction-arm, powered and hydraulic systems: specialist / trained use only, under their specific procedures.
• Do not infer low risk from a small target torque: small fasteners and bits can strip or snap, while high-torque systems can create severe reaction injuries.
• A real "false confidence" trap remains: the tool can be used confidently and wrong, producing a fastener that looks torqued but is not. A first-time user needs: the procedure and units; the certified range; a fully seated correct socket; slow controlled force; stopping at the indication.

revival: the pull is doing a maintenance job right instead of guessing, voiced by Sawmill Creek participants on jointer-knife rotation ("this will provide consistency for each insert") and the plain satisfaction of a correct, felt release in place of an eyeballed guess [TW19].

anatomy

Common designs, not universal anatomy; the mechanism varies by maker and model.

• Drive head / square drive, the socket-mounting end (1/4" to 3/4" and up, matched to socket size). A ratcheting square-drive mechanism, where fitted, is a convenience feature, not the torque-sensing mechanism.
• Handle / beam (tube). On a beam-type wrench this is the sensing element, flexing under Hooke's Law; on an adjustable click-type wrench the outer handle rotates to set the value.
• Torque-setting scale / micrometer collar. On many adjustable click wrenches, a coarse number on the fixed tube plus a fine number on the rotating collar, added together [TW15][TW18].
• Release mechanism (common click designs). In CDI's repair manual for one micrometer click model, a threaded load screw compresses a coil spring against a cam and a small pawl kicks free when applied torque overcomes the spring's preload; the manual stocks pawls in graduated sizes for fine calibration [TW15]. Other designs use a ball in a detent (cam-over) [TW43]. Do not read the internal release mechanism as the same thing as the square-drive ratchet pawl.
• Calibration lock / lock ring, on many adjustable models, a collar that must be released to change the setting and then locked.
• Torque seal / anti-tamper paint. CDI's manual describes a lacquer dab across the calibration joint after factory calibration; a cracked or missing dab is its sign a used wrench may have been tampered with [TW15]. This is the factory's tamper seal, a different thing from a witness mark on a joint.

variations (living)

Each row is a separate system with its own mechanism, setup checks, calibration chain and hazards. They do not inherit one another's safety rating.

trades

Plumbing/pipefitting; structural steel (calibrated-tool-free methods governed by the structural-bolting specification) [TW1][TW2]; automotive and motorcycle repair; aerospace/aviation maintenance (torque striping as standing practice) [TW16][TW17]; oil & gas and marine bolting (hydraulic torque wrenches and manual multipliers, trained use); bicycle mechanics; small-engine repair; and, narrowly but genuinely, woodworking machine maintenance (cutterhead insert screws, per the model's manual).

lineage

• Ancestors: plain wrenches/spanners and the "feel" tradition documented in the Adjustable Wrench entry; on structural steel, the turn-of-nut method (snug the joint, then rotate the nut a specified fraction of a turn) and related calibrated-wrench, direct-tension-indicator and twist-off methods, which the governing structural-bolting specification (not this entry, and not a distributor summary) defines. A hand torque wrench is not the universal way to pretension structural bolts [TW1][TW2].
• Descendants / variations: indicating tools (beam, dial, electronic, the last sometimes adding angle measurement) and setting/signalling and limiting tools (click, cam-over), plus a low-torque precision-driver sub-family (Wera, Wiha) [TW6][TW7]. Manual torque multipliers, pneumatic/electric torque tools and hydraulic torque wrenches are separate industrial classes, not a single lineage of "bigger torque wrenches" [TW3][TW44].
• Contemporaries: Impact Wrench, powered, fast, and by its own entry's sourcing imprecise unless electronically torque-controlled; a torque wrench is the check step that often follows it on safety-critical work, though smart impacts and torque-control power tools can be appropriate under a validated procedure, so manual finishing is not claimed as universal. Adjustable Wrench, no metering mechanism. The torque angle gauge, a separate paired instrument for torque-plus-angle procedures [TW8].

history

Origin: stated cautiously. The earliest primary record verified in this pass is John H. Sharp's US 2,007,880, "Torque-measuring wrench," filed May 18, 1931, published July 9, 1935. Its pointer, pivoted on the housing, moves over a scale that can be calibrated in inch-pounds, so it describes an indicating wrench [TW52]. A second documented branch is US 2,074,079, "Torque Measuring Wrench," filed by Conrad Bahr and George H. Pfefferle on June 12, 1935, granted March 16, 1937, assigned to S.R. Dresser Mfg. Co., with a DATAMP record tying it to the Bonney No. 66 [TW11][TW12]. Sharp, Bahr/Pfefferle and Sturtevant are documented branches, not a clean linear succession.

The Conrad Bahr 1918 story is not established. Tool-history blogs credit Bahr, a New York City Water Department foreman, with the concept around 1918 [TW9][TW10]. This rests on secondary blogs only; no primary record of 1918 was located. The Bahr/Pfefferle patent does not itself verify it. Treat as a story, not a date.

Key evolutions, read through Petroski's "form follows failure" lens:

• Beam type. Paul A. Sturtevant's US 2,283,707 was filed March 28, 1938 and granted May 19, 1942. These are two different patent events, not conflicting accounts. It describes a hand tool with a deflecting steel beam connecting head and handle, a pointer fixed to the head reading against a graduated dial on the handle, to let mechanics tighten fasteners to manufacturer-recommended values instead of by feel [TW14][TW53]. It was not the first indicating wrench (Sharp's pointer-and-scale design is earlier), so "fixed the lack of a mechanical readout" is withdrawn. Secondary sources tie its development to Walter P. Chrysler and Cedar Rapids Engineering Co. [TW9].
• Click / adjustable setting type. The release spring-and-cam arrangement frees the eyes from a dial. Later patent literature states that adjustable click-type wrenches "have been commercially produced and have been in wide use since the 1950s," citing Booth's US 2,704,472 (1955) and Livermont's US 2,732,747 (1956) in its prior-art list [TW44]. An earlier version of this entry credited a 1964 Jon Pape filing with the first adjustable click patent. That "first" is withdrawn; no Pape patent was located and it contradicts the 1950s record.
• Cam-over / break-over. Developed to address over-torque after release by replacing the pawl with a ball held in a detent, per the patent background [TW43].
• Electronic with angle measurement. For torque-plus-angle procedures, a measured rotation angle past an initial torque is part of the joint procedure. Torque-plus-angle is a joint procedure, not a synonym for torque-to-yield, and angle-capable digital tools should be described as such only when the exact model measures angle [TW4][TW5][TW8].
• Multipliers and hydraulic wrenches are covered in the variations table. The earlier line crediting "George Sturdevant in 1968" with the multiplier concept rested on weak trade sourcing and is withdrawn as unverified.

✦ human side

Traditions & trades: plumbing/pipefitting (the Bahr/Pfefferle patent's stated use is pipeline joint couplings); structural steel erection, where turn-of-nut and related methods are governed by the structural-bolting specification [TW1][TW2]; automotive and motorcycle repair; aerospace/aviation maintenance, where torque striping, a paint mark across a torqued fastener and its seat, is standing practice as a later-movement indicator (see Fix 10 note under correct use) [TW16][TW17]; oil & gas and marine bolting, where hydraulic torque wrenches and manual force multipliers are trained-use equipment; bicycle mechanics, a low-torque trade (Park Tool sells three distinct torque-tool types; the entry's earlier "most under 10 Nm" characterization is carried from the original run and not re-read here) [TW18]; and, narrowly but really, woodworking machine maintenance.

Named hands: Conrad Bahr and George Pfefferle (the 1935 patent); John H. Sharp (the earliest primary record verified here); Paul Sturtevant (the beam-type patent). On the woodworking side, a multi-year Sawmill Creek thread on jointer/planer carbide-insert torque names practitioners including John Ziebron, glenn bradley, Peter Christensen, John K Jordan, Ken Fitzgerald and Jesse Busenitz [TW19]. That thread is used here only as a lead to the model-specific manual, never as the authority for a value, sequence or lubrication state, so none of the numbers quoted in it appear in this entry. Two points from it survive as general observations rather than instructions: Christensen's remark that "it isn't the insert that dictates the torque, it is the screw and the metal it's screwed into," and Busenitz's first-person account of using an impact driver on those screws ("wayyyyy too much torque... I need to order some spare cutters"), a woodworking-specific instance of the impact-then-hand-torque failure the Impact Wrench entry documents. A staged procedure re-quoted by Ken Fitzgerald across two threads is a forum practice, not a standard; whether a given cutterhead wants a staged pass is the cutterhead manual's call.

Stories: a 2015 Garage Journal thread, "Before the torque wrench was invented," carries practitioner testimony on how the trade managed tightening before the tool existed: folk heuristics like "snug, tight, very tight," and a named former bridge-crew worker's account of structural bolts tightened by modified impact guns with a counted turn-fraction, using massive forged spud wrenches [TW1]. This is forum testimony, not a procedure.

✦ proof of existence

• Earliest primary patent records verified here: US 2,007,880 (John H. Sharp, filed May 18, 1931, published July 9, 1935), an indicating wrench with pointer and scale, read directly on Google Patents [TW52]. US 2,074,079 (Bahr & Pfefferle, filed June 12, 1935, granted March 16, 1937), the commercially documented branch, manufactured as the Bonney No. 66 [TW11][TW12].
• Museum/collection records: the Smithsonian National Air and Space Museum holds a Space Shuttle-flown torque wrench (object nasm_A20000670000), confirmed by search-result title only and not fetched in full; a lead for a future pass, not a full citation [TW40].
• Patents: US 2,074,079 (Bahr/Pfefferle, 1937); US 2,007,880 (Sharp, 1935); US 2,283,707 (Sturtevant, filed 1938, granted 1942, no date conflict); US 2,704,472 (Booth, 1955) and US 2,732,747 (Livermont, 1956), named as click-wrench prior art in US 5,662,012's background [TW44]; US 2,196,223 (Montgomery, 1940); US 2,312,104 (Larson, 1943); US 4,485,703 (Grabovac & Vuceta, 1984, ball-detent click mechanism, secondary citation) [TW11][TW12][TW14][TW9][TW41].
• Archaeological/documentary evidence: none applicable; a documented 20th-century invention whose ancestor is the plain wrench and the "feel" and turn-of-nut traditions.

references

Sources added in this fix pass are TW43 to TW54. Source quality note per the reviewer: patents, standards pages, manufacturer guidance and IMCA reports are strong; blogs, Wikipedia, forums, retailer pages and search snippets are discovery leads and carry no claim of invention priority, universal torque values, calibration intervals or safety limits.

The reviewer's corrections (12)

• BLOCKER — 1. Separate indicating, setting, and limiting tools — Exact claim: The torque wrench “measures and limits” torque, exists to “report that number,” and applies and signals a specified amount.
• BLOCKER — 2. Remove the universal 50%-then-100% star/spiral sequence — Exact claim: On any multi-fastener assembly, use a star or spiral order, first at roughly half torque and then at final torque.
• BLOCKER — 3. Rewrite the physical-risk assessment — Exact claim: Torque-wrench work has low physical-injury risk and lacks socket-shattering, jaw-slip, or knuckle-strike hazards.
• BLOCKER — 4. Split multipliers, powered tools, and hydraulic wrenches — Exact claim: A single “hydraulic/pneumatic torque multiplier” lineage and variant removed the operator's strength ceiling.
• SHOULD — 5. Correct the calibration-interval attribution — Exact claim: ISO 6789 recommends calibration annually or every 5,000 cycles.
• SHOULD — 6. Remove the universal “below 20%” rule — Exact claim: Torque wrenches are inaccurate below roughly 20% of their range and may fail to click.
• SHOULD — 7. Replace the generic anti-seize percentage — Exact claim: Anti-seize can create 30–65% more clamp force at the same torque.
• SHOULD — 8. Correct the patent chronology and claims of primacy — Exact claim: Sturtevant has a 1938/1942 date conflict and fixed the lack of mechanical readout; Jon Pape's 1964 filing was the first adjustable click patent; Conrad Bahr's 1918 invention is treated as established history.
• SHOULD — 9. Make storage and force-direction guidance model-specific — Exact claim: Store every click wrench at the lowest nonzero setting; never push, only pull.
• SHOULD — 10. Fix torque-marking language — Exact claim: Torque striping functions as proof that a fastener was correctly torqued or remains at specification.
• SHOULD — 11. Replace forum and search-synthesis values with direct manuals — Exact claim: Jointer insert screws commonly use 45–55 in-lb; SawStop's arbor nut is “snug, not spec'd”; cutterhead sequencing is generalized from forum discussion.
• SHOULD — 12. Remove class-wide accuracy and mechanism claims — Exact claim: Digital tools are ±2%, dial tools ±4%, and click tools share a spring-loaded pawl/cam mechanism.