A hook-tooth escapement is a clockwork escapement in which the escape-wheel teeth are formed as forward-curved hooks that engage flat or shallow pallets to release the gear train one tooth at a time. Antique turret-clock and early longcase restorers still meet it regularly. The hook geometry catches the pallet on locking and slides along the impulse face as the pendulum swings, transferring a small kick of energy each beat. The result is a robust, low-precision escapement that runs reliably on coarse train tolerances — typically ±30 seconds per day in a good 18th-century turret movement.
Hook-tooth Escapement Interactive Calculator
Vary the pendulum beat interval and escape-wheel tooth count to see pendulum length, beat rate, wheel speed, and tooth advance.
Equation Used
The calculator uses the simple-pendulum timing relation. A 1-second pendulum has a 1 second beat interval and a 2 second full period, giving about 994 mm length at standard gravity. The escape wheel is assumed to advance one tooth per beat, so wheel speed depends on beat rate and tooth count.
- Simple pendulum, small swing angle.
- Standard gravity is g = 9.81 m/s^2.
- One escape-wheel tooth is released per beat.
- Wheel tooth count is treated as an integer slider value.
How the Hook-tooth Escapement Actually Works
The hook-tooth escapement works by letting one curved tooth of the escape wheel drop onto a pallet, push against it through a short impulse arc, then release so the next hook can land. Picture a row of forward-leaning hooks rather than the radial teeth you see on a deadbeat escape wheel — that forward rake is what gives the geometry its name and its behaviour. The pallet sits in the path of the hook tip, and as the pendulum (or balance) swings through its centre position, the hook slides along the pallet's impulse face and delivers a small push of energy before unlocking.
The design exists because early clockmakers needed an escapement that tolerated rough pivots, soft iron arbors, and uneven wheel cutting. The hook profile is forgiving — if the lift angle drifts a degree or two, or if a pallet face wears 0.1 mm shallow, the clock keeps ticking. Compare that to a deadbeat escapement, where a 0.05 mm error on the locking face can stop the clock dead. The hook-tooth gives up precision in exchange for that tolerance, and it gives recoil — the escape wheel actually swings backward a few arc-minutes after lock, because the hook's curved back drives the pallet rearwards as the pendulum overshoots.
If the impulse face geometry goes wrong you'll see specific symptoms. Too steep a lift angle and the pendulum gets thumped, amplitude climbs, but the rate becomes erratic with circular error. Too shallow and the clock loses energy to drop loss between teeth and stops on cold mornings when oil thickens. Worn hook tips — the most common failure mode in 200-year-old movements — produce a double-tick on locking because the tooth lands, slides, and lands again before settling. You diagnose this by ear, then confirm with a 10× loupe looking for a rounded or chipped hook profile against a sharp original.
Key Components
- Escape Wheel with Hooked Teeth: A wheel typically of 30 teeth in turret work, each tooth forged or filed into a forward-curving hook with a tip radius of roughly 0.3-0.5 mm on a 150 mm wheel. The hook's leading face is the impulse-delivery surface; the trailing curve is what allows recoil. Tooth-to-tooth pitch error must stay under 0.1 mm or you'll hear an audible limp in the beat.
- Pallets (Verge or Anchor Style): Two flat or shallow-curved pallet faces mounted on a rocking arbor, spaced to span the correct number of teeth — usually around 7 teeth in an anchor configuration. The impulse face runs at a lift angle of 3° to 5° depending on the wheel diameter and pendulum amplitude required.
- Pallet Arbor and Crutch: The arbor carrying the pallets transmits motion to the pendulum through a crutch — a slotted lever engaging a pin on the pendulum rod. Side play in the arbor pivots must stay under 0.05 mm or the pallet faces drift out of plane and engagement becomes intermittent.
- Pendulum or Foliot: Provides the time-keeping reference. On a 1-second pendulum the rod is approximately 994 mm from pivot to centre of bob mass. The hook-tooth escapement supplies just enough impulse — typically 2-4 mJ per beat — to maintain the swing against air drag and pivot friction.
- Impulse Face: The flat or near-flat surface on the pallet that the hook tip slides along during impulse. Surface finish matters: an Ra of 0.4 µm or better is what original makers achieved with hand stoning, and a rougher finish doubles friction losses and accelerates wear on the hook tip.
Where the Hook-tooth Escapement Is Used
You meet the hook-tooth escapement almost exclusively in restoration and heritage work. It's a survivor mechanism — nobody designs new ones, but plenty of working examples exist in church towers, longcase clocks, and early stable-clocks across Europe. Wherever you find a 17th- or 18th-century iron clockwork movement still running, there's a fair chance the escapement is some variant of hook-tooth.
- Heritage Horology: Restoration of 17th-century lantern clocks where the original verge-and-foliot was later converted to a hook-tooth anchor by a country clockmaker — common in surviving Thomas Tompion-era English provincial work.
- Turret Clock Conservation: Operating turret clocks such as the Salisbury Cathedral clock variants and surviving Smith of Derby tower movements where hook-tooth wheels still drive the going train.
- Stable and Estate Clocks: Early-1800s English stable-yard clocks by makers like John Moore of Clerkenwell, where the hook-tooth wheel and anchor pallet combination runs an exposed weight-driven movement.
- Museum Demonstration Movements: Cutaway working models in collections such as the British Museum's clockwork gallery, where hook-tooth geometry is shown because the impulse and recoil are visually obvious to the public.
- Educational Clockmaking: Teaching workshops at the British Horological Institute and West Dean College use hook-tooth escapements as a first-build project because the geometry tolerates beginner filing errors.
- Auction House Authentication: Specialist horology desks at houses like Bonhams and Dreweatts inspect escape-wheel tooth form to date provincial English movements — hook profile and tooth count are dating fingerprints.
The Formula Behind the Hook-tooth Escapement
The energy delivered to the pendulum each beat is what determines whether the clock will start, run, and survive a winter. You compute it from the driving torque at the escape-wheel arbor, the lift angle of the pallet, and the angular distance the wheel rotates during impulse. At the low end of the typical operating range — say 0.5 N·mm of arbor torque on a small bracket clock conversion — you're delivering around 1 mJ per beat, just enough to keep a light pendulum alive. At the nominal turret-clock value of 5 N·mm torque you're at roughly 10 mJ, comfortable margin. Push above 20 N·mm in a heavy tower movement and the impulse becomes violent, amplitude climbs past 4°, and circular error swamps the rate.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Eimpulse | Energy delivered to the pendulum per beat | J (joules) | ft·lbf |
| Tarbor | Torque at the escape-wheel arbor under the maintaining force from the train | N·m | lbf·in |
| θlift | Lift angle through which the escape wheel rotates during impulse | degrees | degrees |
| π / 180 | Conversion factor from degrees to radians | rad/deg | rad/deg |
Worked Example: Hook-tooth Escapement in an 1780s English provincial turret clock recommissioning
A heritage estate in the Scottish Borders is recommissioning an 1780s provincial turret clock attributed to a Hawick maker, with a hook-tooth escape wheel of 150 mm diameter, 30 teeth, driving a 1.5-second pendulum from a 25 kg driving weight on a 100 mm winding barrel. You measure a lift angle of 4.5° on the pallets and need to confirm the impulse energy per beat is in the right range to keep the pendulum at 2.5° amplitude through a Borders winter.
Given
- Driving weight = 25 kg
- Barrel radius = 50 mm
- Train multiplication to escape arbor = 1:96 ratio
- θlift = 4.5 degrees
- Pendulum period = 3.0 s (1.5 s pendulum, full swing)
Solution
Step 1 — compute the torque at the great wheel arbor from the driving weight and barrel radius:
Step 2 — divide by the train ratio to get torque at the escape arbor (ignoring train friction for this first pass; in real movements you lose 30-40% to pivot friction):
Apply a realistic 35% friction loss across the train and you land at roughly 83 N·mm at the escape arbor. That's a healthy turret-clock figure.
Step 3 — convert lift angle to radians and compute impulse energy at the nominal 4.5° lift:
That 6.5 mJ per beat is squarely in the sweet spot — enough to maintain 2.5° amplitude on a heavy 1.5-second pendulum with the bob mass typical of turret work, with margin for cold-oil mornings.
Now bracket the operating range. At the low end, suppose the train friction has worsened to 60% loss (worn pivots, oxidised pinions) and effective arbor torque drops to 51 N·mm:
At 4 mJ the clock will still run but amplitude collapses to around 1.5° and rate becomes weather-sensitive — exactly the symptom an estate manager describes as "it stops on the coldest nights". At the high end, if a previous restorer over-reshaped the pallets to a 7° lift angle:
10 mJ slams the pendulum to 4°+ amplitude, circular error climbs into the seconds-per-day range, and you'll hear a hard double-tick on the locking face. Neither extreme is acceptable for accurate timekeeping.
Result
The nominal impulse energy comes out at 6. 5 mJ per beat — the figure you want to see on a healthy turret movement of this size. That number means the pendulum will hold 2.5° amplitude steadily and the clock should keep within ±20 seconds per day across a typical British year. Compare the range: 4 mJ at the worn-train low end is survival mode, 6.5 mJ nominal is comfortable, 10 mJ at the over-lifted high end is destructive. If you measure the impulse energy or pendulum amplitude well below predicted, the most likely causes are: (1) pinion leaves worn into a hook profile rather than involute, costing 20-30% of train torque on every transfer, (2) the crutch pin running in a sloppy slot greater than 0.3 mm clearance, eating impulse to lost motion, or (3) a bent pallet arbor that lets one pallet engage 0.2 mm deeper than the other, producing alternating strong and weak beats.
Choosing the Hook-tooth Escapement: Pros and Cons
The hook-tooth sits between the verge and the deadbeat in horological history, and the engineering tradeoffs reflect that middle position. You pick it (or accept it, in restoration) because of what it tolerates, not because of what it achieves.
| Property | Hook-tooth Escapement | Deadbeat Escapement | Verge Escapement |
|---|---|---|---|
| Typical accuracy (seconds per day) | ±20 to ±60 s/day | ±2 to ±10 s/day | ±60 to ±300 s/day |
| Lift angle range | 3° to 5° | 1° to 3° | 30° to 50° |
| Tolerance to pivot wear | High — runs with 0.1 mm pivot slop | Low — stops at 0.03 mm pivot slop | Very high — runs with 0.3 mm slop |
| Recoil behaviour | Yes, mild recoil | No recoil (deadbeat) | Yes, strong recoil |
| Surface finish required on impulse face | Ra ≤ 0.4 µm | Ra ≤ 0.2 µm | Ra ≤ 1.6 µm acceptable |
| Suitable application | 18th-19th century turret and longcase work | Precision regulators and observatory clocks | Pre-1670 lantern clocks and rough movements |
| Restoration cost (UK rates, 2024) | £800 to £2,500 per escapement | £1,500 to £4,000 | £500 to £1,500 |
Frequently Asked Questions About Hook-tooth Escapement
Cold-stop on a hook-tooth almost always traces to impulse energy sitting too close to the minimum survival threshold. At 20°C the lubricant is fluid, train friction is low, and you're delivering enough joules per beat to maintain swing. Drop to 5°C and the oil viscosity on the pivots can triple, train torque at the escape arbor falls 20-30%, and the impulse drops below what the pendulum needs to overcome air drag and pivot friction.
The diagnostic is straightforward: warm the movement back up and time how long it runs before stopping at low temperature. If it runs for 6 hours and dies, you're marginal. The fix is rarely the escapement itself — it's worn pivots, wrong oil grade, or a train pinion that's lost its polish. Re-jewel or re-polish the going-train pivots and switch to a low-viscosity clock oil like Moebius 8030 before you touch the pallets.
Choose hook-tooth when the movement is being conserved as a historical artefact, when the going train was originally cut to suit the higher impulse and recoil of a hook-tooth, and when the timekeeping target is realistic — anything tighter than ±15 s/day is the wrong specification for this escapement.
Convert to a deadbeat (Graham-style) only if the client has accepted that originality is being sacrificed for accuracy, and if the great wheel and pinion train can be re-cut or re-tuned for the lower impulse demand. A deadbeat dropped into a train sized for hook-tooth torque will hammer the pallets and wear them within five years. The decision is a horological one, not a purely engineering one — get it in writing from the client either way.
0.4 mm pitch error on a 150 mm wheel is roughly 5× the limit you want for clean beat. You'll hear it as a limping tick — alternating loud-soft-loud-soft beats — and the rate will wander because each beat delivers a different impulse energy.
Dressing is possible if the error is concentrated in 1-3 teeth and the rest of the wheel is within 0.1 mm. Identify the bad teeth with a depth gauge against a master template, then dress the leading face (not the hook tip) to bring the engagement timing back. If the error is distributed across most teeth, you're looking at re-cutting the wheel — original 18th-century iron wheels are usually worth saving with new teeth silver-soldered into milled slots, but that's specialist work. Don't shorten the hooks to compensate; you'll cripple the impulse delivery.
Hook-tooth escapements have meaningful circular error because the impulse arc is large compared to a deadbeat. If you move the clock to a slightly out-of-level mantel or a softer floor that absorbs vibration differently, the pendulum amplitude changes — and amplitude change on a circular-error-sensitive escapement directly shifts the rate.
Check level first with a precision spirit level on the seatboard, not the case. 0.5° out of level can shift amplitude by 0.3° and rate by 20-40 s/day. Second, listen for vibration coupling: a wooden floor flexing under footfall feeds back into the suspension and modulates amplitude. The fix is mechanical isolation, not regulation — re-regulate only after the clock is properly mounted.
Asymmetric pallet wear is almost always a beat error, not a material defect. If the pallets are not symmetrically placed about the escape-wheel centre line, one pallet receives the hook impact slightly off-centre or at a steeper effective angle than the other, and that pallet wears faster.
Set the clock in beat audibly first — equal interval between tick and tock at zero amplitude error. Then check pallet geometry against the wheel: scribe a centre line through the escape-wheel arbor and the pallet arbor and confirm both pallets contact at equal angles. A 0.5 mm displacement of the crutch pin or a bent pallet arm will quietly destroy a pallet face over a few years of running. Fix the geometry before you re-stone the worn face, otherwise you'll be back at the same wear pattern within 18 months.
Three tells, in order of reliability. First, tooth-cutting marks: original 18th-century wheels were filed by hand and show subtle pitch variation of 0.05-0.15 mm and individual file-stroke marks under a 10× loupe. Victorian replacements were machine-cut and show consistent pitch within 0.02 mm and uniform tool marks.
Second, the iron itself: pre-1820 wrought iron has a visible fibrous grain structure when etched, while later Bessemer or wrought-equivalent steel is uniform. Third, hook profile: 18th-century provincial work has a rounder, less aggressive hook because the maker filed it to feel rather than to a template — Victorian hooks are sharper and more uniform tooth-to-tooth. Combine all three observations before you commit to a date — any one alone can mislead.
References & Further Reading
- Wikipedia contributors. Escapement. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
