The Anchor Escapement is a pendulum-driven escapement that uses a pivoted, anchor-shaped lever with two pallets to release one tooth of the escape wheel per pendulum swing. Typical accuracy sits at 1-2 minutes per week in a domestic longcase clock, with escape wheels running 30 teeth at 0.5-second beats. It replaced the verge to allow longer, slower pendulums and tighter rates. William Clement's 1670s longcase clocks were the first widespread application, and the design still drives most surviving Georgian and Victorian wall and floor clocks today.
Anchor Escapement Interactive Calculator
Vary escape-wheel teeth, pendulum period, swing arc, and pallet span to see beat timing, wheel advance, speed, and pallet coverage.
Equation Used
The calculator uses the anchor escapement timing rule that one wheel tooth is released on each beat. A 30-tooth escape wheel therefore advances 360/30 = 12 degrees per beat. Wheel speed follows from the beat interval and number of teeth.
- One escape-wheel tooth is released per pendulum beat.
- Pendulum period T is the full left-right-left cycle, so beat interval is T/2.
- Ideal escapement timing is calculated; recoil losses, friction, and drop errors are not included.
How the Anchor Escapement Actually Works
The Anchor Escapement, also called the Recoil Escapement when its pallets push the escape wheel briefly backwards on each beat, works by letting one tooth of the escape wheel slip past at every swing of the pendulum. The anchor itself is a steel arm pivoted on the same axis as the pendulum suspension or driven by a crutch from it. Two pallet faces sit at the ends of the anchor — entry and exit — and they alternately catch and release teeth of the escape wheel. Drive torque comes from the going train, ultimately the weight or mainspring, and a small pulse of that torque transfers to the pendulum on every release. That pulse is what keeps the pendulum swinging against air drag and pivot friction.
Geometry is everything here. The pallet faces are angled so that when a tooth lands on them, the continued swing of the pendulum first pushes the escape wheel slightly backwards — the recoil — before the tooth slides off the pallet face and the wheel jumps forward by half a tooth. That recoil is why the second hand of an old longcase clock visibly twitches backwards between ticks. The pallet span typically covers 7.5 teeth on a 30-tooth wheel, and the impulse angle on the pallet face usually sits between 2° and 4° measured at the anchor pivot. If you cut that angle wrong by even half a degree the clock will either trip (run without engaging properly) or stall under low mainspring torque near the end of its run.
If the pallet faces wear, the drop — the small free travel of the wheel before a tooth catches — opens up. Excessive drop wastes drive torque as noise and shock, and you'll hear it as a loud, uneven tick. Too little drop and the wheel won't release cleanly, and the clock stops. Bent crutch wires, dry pivot holes, and out-of-beat suspensions are the three failures you'll see most often when servicing a 200-year-old movement.
Key Components
- Escape Wheel: A thin brass wheel with 30 pointed, undercut teeth in most longcase work. It delivers torque to the pallets and must run true to within about 0.05 mm radial — any more and one side of the pallet sees more drop than the other, throwing the clock out of beat.
- Anchor (Pallet Arm): A pivoted steel lever shaped like a ship's anchor, carrying the entry and exit pallet faces. Pallet span typically equals 7.5 teeth on a 30-tooth wheel. The anchor pivots in jewelled or polished brass holes with about 0.02 mm clearance.
- Pallets: Hardened steel faces ground to a 2-4° impulse angle. They take the full impact of each tooth landing thousands of times per day, so the working face is polished to a mirror to keep friction predictable. Worn pallets cause stalling under low torque.
- Crutch: A light wire or fork that links the anchor to the pendulum rod, isolating the pendulum from sideways forces. The crutch must be straight and free in the pendulum slot — bind here and the pendulum loses amplitude within minutes.
- Pendulum: Sets the timekeeping rate. A seconds pendulum is 0.994 m to the centre of oscillation. The Anchor Escapement only works well with pendulum amplitudes of 4-6° total swing — push higher and circular error dominates the rate.
Who Uses the Anchor Escapement
The Anchor and Lever Escapement, sometimes called the Rocking Escapement because of the visible rocking motion of the anchor, sits inside the vast majority of weight-driven and spring-driven mechanical clocks built between 1670 and the rise of the deadbeat in serious regulators. It is cheap to make, forgiving on torque variation, and runs for centuries with light maintenance. You'll find it in everything from 17-shilling cottage clocks to museum-grade longcase movements.
- Horology — Domestic Clocks: Standard longcase (grandfather) clocks by makers like Thomas Tompion and later by Comitti of London — the Recoil pendulum escapement drives a 0.5-second or 1-second beat with a 30-tooth escape wheel.
- Horology — Wall Clocks: Vienna regulators, English dial clocks, and German Black Forest wall clocks all use the Anchor Escapement with shorter pendulums in the 0.4-0.5 second range.
- Horology — Bracket Clocks: English bracket clocks and French mantel clocks built between 1700 and 1900, where compact size and tolerance to bumping made the recoil action preferable to the deadbeat.
- Education — Mechanical Engineering: Classroom escapement demonstrators sold by Tamiya and laser-cut acrylic kits used in MIT 2.007 and similar mechanism courses to teach impulse and timekeeping.
- Restoration & Conservation: Conservation work on heritage tower clocks, such as parish church turret movements predating the Denison gravity escapement, where the original Anchor Escapement is rebuilt rather than replaced.
- Hobbyist Clockmaking: Wooden gear clocks built from Brian Law's plans (Clayton, Genevieve) — most use a wooden anchor running on a wooden 40-tooth escape wheel, demonstrating the geometry without needing a machine shop.
The Formula Behind the Anchor Escapement
The core relationship for an Anchor Escapement ties the pendulum period to the escape wheel rotation, which sets how the second hand advances. At the low end of the typical operating range — a 2-second pendulum on a longcase clock — the escape wheel turns once every minute, and the seconds hand advances in 2-second jumps with visible recoil. At the nominal 1-second pendulum the wheel turns twice per minute and the hand ticks every second. At the high end — a 0.4-second Vienna regulator pendulum — the wheel spins five times faster and recoil becomes visually invisible but mechanically louder. Knowing the period and tooth count tells you exactly how to lay out the going train above the escape wheel.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Twheel | Time for one full revolution of the escape wheel | seconds | seconds |
| Nteeth | Number of teeth on the escape wheel | count | count |
| Tpendulum | Period of one complete pendulum swing (out and back) | seconds | seconds |
| Lp | Pendulum length to centre of oscillation | metres | inches |
| g | Gravitational acceleration | m/s² | ft/s² |
Worked Example: Anchor Escapement in a Victorian English dial wall clock
You are restoring a Victorian English dial wall clock and need to specify the escape wheel and pendulum so the minute hand advances correctly. The movement has a 30-tooth escape wheel and you want the wheel to complete one revolution every 30 seconds, which means the centre seconds wheel above it can be geared 1:1. Pick the pendulum period and length, then check what happens if you misjudge the pendulum length at the build stage.
Given
- Nteeth = 30 teeth
- Twheel = 30 seconds
- g = 9.81 m/s²
Solution
Step 1 — at the nominal target, solve for pendulum period from the escape geometry:
A 2-second period means a seconds pendulum — the half-swing takes 1 second, and one tooth escapes per second. This is the classic longcase rate.
Step 2 — back-calculate the pendulum length needed for that period:
That is the textbook 994 mm seconds pendulum length. Build it to within ±2 mm and you can rate-correct the rest at the rating nut.
Step 3 — at the low end of the typical operating range for this size of dial clock, suppose the builder shortens the pendulum to 0.621 m for a 1.58-second period (a common compact wall clock spec):
The escape wheel now spins faster — once every 23.7 seconds — so the seconds hand ticks more often than once per second and any 1:1 centre seconds gearing is wrong. You'd need a different motion-work ratio.
Step 4 — at the high end of typical, a 1.2 m pendulum for a tall regulator gives:
The wheel turns slower than the desired 30 s, the clock loses 6 minutes per hour, and no amount of rating-nut adjustment will fix it — you'd have to shorten the pendulum back toward 0.994 m.
Result
The nominal answer is a 2. 0-second seconds pendulum at 0.994 m driving a 30-tooth escape wheel that completes one revolution every 30 seconds. In practice this gives the familiar slow, deliberate tick of an English dial clock — every tick is one second of real time, with a small visible recoil twitch on each beat. At the low end with a 0.621 m pendulum the wheel spins in 23.7 s and the going train above must be re-geared; at the high end with a 1.2 m pendulum it slows to 33.0 s and loses time uncorrectably. If your finished clock measures wheel rotation slower or faster than 30 s ±0.5 s, suspect (1) a pendulum bob clamped at the wrong height on a metric vs imperial rod, (2) a crutch that is binding on the pendulum slot and damping amplitude below 3°, or (3) escape wheel teeth that have been incorrectly stoned during cleaning, changing the effective tooth pitch.
When to Use a Anchor Escapement and When Not To
Choosing between an Anchor Escapement and its alternatives comes down to accuracy targets, drive torque variation, and how much you care about recoil. The classic Anchor (the Recoil Escapement) is forgiving but inherently introduces rate variation with torque. The deadbeat eliminates recoil at the cost of tighter geometry. The grasshopper trades almost-zero friction for fragility. Pick by application, not by reputation.
| Property | Anchor Escapement (Recoil) | Deadbeat Escapement | Grasshopper Escapement |
|---|---|---|---|
| Typical accuracy (domestic build) | 1-2 min/week | 5-15 sec/week | 1-5 sec/week |
| Sensitivity to drive torque variation | High — rate changes with mainspring state | Low — torque isolated from pendulum | Very low |
| Recoil action | Yes, visible on seconds hand | No, dead drop | No, but unusual impulse path |
| Manufacturing complexity | Low — forgiving pallet angles | Medium — tight pallet geometry needed | High — many pivoted parts |
| Service life between overhauls | 30-50 years typical | 20-40 years (pallets jewelled) | 10-20 years (delicate) |
| Maximum recommended pendulum amplitude | 6° total swing | 3° total swing | 1.5° total swing |
| Best application fit | Domestic longcase, wall, bracket clocks | Regulators, observatory clocks | Precision regulators, museum pieces |
Frequently Asked Questions About Anchor Escapement
This is the classic torque-dependence of the Recoil Escapement. When the weight is high or the mainspring is fully wound, drive torque is greater, the impulse delivered to the pendulum is larger, and pendulum amplitude rises. Larger amplitude means circular error pushes the rate slightly slower in theory, but the recoil action also increases, which actually causes the clock to gain because the impulse profile shifts.
You cannot fully eliminate this — it is intrinsic to the design. The fix is either a remontoire to deliver constant torque, or upgrade to a deadbeat. For a domestic clock, accept ±30 seconds drift across the week as normal.
Watch the seconds hand. If it visibly jumps backwards a small amount on each tick before snapping forward, you have an Anchor Escapement — that backward motion is the recoil. If the hand simply pauses dead between ticks with no backward motion, it is a deadbeat.
You can also listen. Recoil escapements have a softer, slightly muddier tick because the wheel is being pushed against the pallet during recoil. Deadbeats sound crisper and more metallic.
Pick the Anchor every time for a wooden build. The deadbeat needs pallet faces held to within roughly 0.05 mm of nominal angle, which wood cannot hold across a humidity cycle. The Anchor's 2-4° impulse angle is forgiving — even a 0.5° error just changes amplitude slightly without stopping the clock.
Brian Law's wooden clock plans all use the Anchor Escapement for exactly this reason, and they run for years on plywood pallets.
An Anchor Escapement needs at least 3° of half-amplitude to reliably clear the pallet faces and accept the impulse. Below that the pendulum runs out of energy before the next tooth releases.
The cause is almost always drag, not the escapement itself. Check the suspension spring for a kink — even a 0.1 mm bend halves amplitude. Then check the crutch fork for tight contact with the pendulum rod. Finally, dirty pivot holes in the going train rob torque before it ever reaches the escape wheel. A full clean and oil normally restores 5° amplitude.
The rate-vs-length relationship is not linear — period scales with the square root of length. A 1% length change gives only a 0.5% rate change, so to gain 30 seconds per day you only need to shorten by about 0.7 mm on a seconds pendulum. Beyond a few millimetres of correction you are usually fighting a different problem entirely.
Common real causes: bent crutch, bob slipped on the rod, or a temperature-induced length change in a steel-rod pendulum. Wood and Invar rods are stable; plain steel changes about 11 ppm per °C, which is 11 seconds per day per 10°C swing.
Yes — they are the same mechanism. "Anchor Escapement" describes the shape of the pallet arm. "Recoil Escapement" describes the action: the brief backward push of the escape wheel on each beat. Some texts also call it the Anchor and Lever Escapement or Rocking Escapement.
All four names describe the design Robert Hooke and William Clement developed in the 1670s. Strictly, a non-recoil anchor (the deadbeat by Graham, 1715) is a separate design — so when someone says "recoil" they specifically mean the original 1670s form.
Radial runout on the escape wheel must stay under about 0.05 mm for a 30-tooth, 50 mm diameter wheel. Above that, one side of the wheel hits the pallet harder than the other, which biases the pendulum to one side and puts the clock out of beat — you'll hear an uneven tick-TOCK rather than a steady tick-tock.
Endshake should sit between 0.05 mm and 0.15 mm. Less than that and the wheel binds when the plates flex under load; more than that and the wheel walks axially, scuffing the pallets unevenly. A simple feeler-gauge check at service time catches this.
References & Further Reading
- Wikipedia contributors. Anchor escapement. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
