An endless cord-winding device for clocks is a closed loop of cord, chain, or gut passed over two pulleys with a ratcheted driving wheel, letting you wind a weight-driven clock without ever disconnecting drive to the movement. The system carries the full driving weight — typically 4 to 12 kg on an 8-day longcase — plus a small counterweight that tensions the loop. It solves the dead-spot problem during winding, where the going train would otherwise stop or reverse. Christiaan Huygens published the design in 1664, and it remains standard practice on quality longcase and turret clocks today.
Endless Cord-winding Device for Clocks Interactive Calculator
Vary the driving weight, counterweight fraction, and pulley diameter to see the continuous torque delivered through the endless cord loop.
Equation Used
The main weight and smaller counterweight create different cord tensions on the two sides of the great-wheel pulley. The available continuous arbor torque is the net weight force, (m - mc)g, multiplied by the effective pulley radius.
- Counterweight percentage is based on the main driving mass.
- Available drive torque is the net cord tension difference times pulley radius.
- Pulley diameter is the effective cord pitch diameter.
- Friction, ratchet losses, and cord stretch are neglected.
How the Endless Cord-winding Device for Clocks Actually Works
The mechanism is a closed loop. The cord runs from the driving weight, up over a grooved pulley fixed to the great-wheel arbor, across to a second idler pulley, and back down to a small counterweight — typically 5 to 10% of the main weight. The great-wheel pulley carries a ratchet and click, so the loop can rotate freely in the winding direction but locks into the wheel when the main weight is doing work. Pull down on the counterweight side and you lift the main weight without ever taking torque off the going train.
Why design it this way? On a normal weight-driven clock with a single rope wrapped on a barrel, winding lifts the weight by reversing the barrel — and during those 30 to 60 seconds, the train has no driving force. Pendulum amplitude drops, the escapement may unlock, and a precision regulator can lose seconds or stop. The endless loop keeps continuous torque on the great wheel because the click stays engaged on the ratchet whichever way you pull. This is the original maintaining power solution, predating Harrison's spring-loaded going-barrel maintaining power by nearly a century.
Tolerances matter. The pulley grooves must match the cord diameter within roughly ±0.2 mm — too loose and the cord slips under load, too tight and it binds when the loop transitions through the groove. The ratchet click must drop fully into each tooth before the wheel reverses direction, so click spring force needs to be light enough to lift cleanly during winding but firm enough to seat under the full weight torque. Common failure modes are click bounce (the click skips a tooth under shock load and the weight runs down freely), cord stretch (gut elongates 2 to 5% over its life and the loop sags off the idler pulley), and groove wear that lets the cord ride high and slip during the running cycle.
Key Components
- Great-wheel pulley: Grooved pulley fixed to the great-wheel arbor, typically 60 to 120 mm diameter on a longcase clock. Drives the going train through the ratchet and click. Groove profile is usually a half-round matched to cord diameter within ±0.2 mm.
- Ratchet and click: One-way locking pair mounted on the great wheel. The click engages teeth cut into a ratchet ring on the pulley so the loop can rotate freely during winding but locks under driving-weight torque. Click spring is sized to give roughly 0.2 to 0.5 N seating force.
- Idler pulley: Free-running pulley mounted in the case top or seatboard, opposite the great-wheel pulley. Carries the loop across to the counterweight side. Must run on a low-friction bearing — pivot friction here directly steals torque from the going train.
- Driving weight: The main weight that powers the going train, typically 4 to 12 kg on an 8-day longcase, hung on a hook or pulley block on one side of the loop. Provides the steady torque that keeps the pendulum swinging.
- Counterweight: Small weight on the opposite side of the loop, typically 5 to 10% of the driving weight. Its job is only to keep the loop taut against the pulleys — too heavy and it steals torque, too light and the cord jumps the groove.
- Cord, gut, or chain: The loop itself. Traditional gut on quality clocks (1.2 to 2.0 mm diameter), modern braided polyester on replicas, or roller chain on turret clocks where load exceeds 50 kg. Cord must be spliced into a continuous loop with no knot — knots will jam the pulley groove.
Who Uses the Endless Cord-winding Device for Clocks
You see this mechanism wherever a weight-driven clock needs to keep precise time through the winding cycle. Spring-driven movements solved the same problem differently with the going barrel, but for any clock running on a falling weight — longcase, turret, regulator — the endless loop is still the cleanest way to wind without losing time. It also shows up in non-clock contexts where continuous drive during reset matters.
- Domestic horology: Quality 8-day English longcase clocks from makers like Thomas Tompion and George Graham routinely used endless gut loops on their month-going regulators. The 1675 Tompion year-going regulator at the Royal Observatory Greenwich used an extended version of the same principle.
- Public clocks: Turret clocks with heavy driving weights — the Smith of Derby movements in mid-Victorian church towers commonly used endless chain loops with cast-iron weights of 50 to 100 kg, wound weekly through a hand crank on the loop side.
- Astronomical regulators: Precision observatory regulators where any winding interruption corrupts the rate measurement. The Shelton regulators used by Cook on his Pacific voyages relied on continuous-drive winding to preserve transit timing accuracy.
- Museum and replica clockmaking: Working replicas of Huygens' 1656 pendulum clock built for science museums — the Boerhaave museum in Leiden displays one — use the original endless-loop arrangement Huygens described in Horologium Oscillatorium.
- Tower and station clocks: Railway station clocks and town hall installations where the public would notice a stopped second hand during the weekly wind. The Dent movement at the Royal Liverpool Building uses endless chain drive.
- Educational horology: Horology programmes at WOSTEP and the British School of Watchmaking build endless-loop demonstration movements as a standard exercise in maintaining power theory.
The Formula Behind the Endless Cord-winding Device for Clocks
What matters in practice is the torque the loop delivers to the great wheel, because that's what drives the pendulum. The driving torque is the net weight differential across the great-wheel pulley times the pulley radius. At the low end of typical operation — say, a 4 kg driving weight with a 0.4 kg counterweight on a 30 mm-radius pulley — you get just enough torque to run a small bracket regulator. At the high end, a 12 kg weight with a 1 kg counterweight on a 60 mm pulley delivers roughly 8× more, suitable for a longcase with a heavy compound pendulum. The sweet spot for an 8-day domestic longcase sits around 6 kg main, 0.5 kg counter, 40 mm pulley radius. Push the counterweight ratio above 15% and you waste torque; drop it below 3% and the cord skips the groove on the slack side.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Tdrive | Net driving torque applied to the great-wheel arbor | N·m | lbf·in |
| mw | Mass of the main driving weight | kg | lb |
| mc | Mass of the tensioning counterweight | kg | lb |
| g | Gravitational acceleration | m/s² | ft/s² |
| rp | Effective groove radius of the great-wheel pulley | m | in |
Worked Example: Endless Cord-winding Device for Clocks in an 8-day longcase regulator restoration
A regional auction house in Bath sends you an 1810 John Roger Arnold longcase regulator with a missing weight system. The great-wheel pulley measures 80 mm diameter (40 mm groove radius), and the original specification lists a 6 kg lead driving weight. You need to size the counterweight and confirm the resulting torque at the great wheel will sustain a 1-second seconds-pendulum with roughly 3° amplitude through the weekly wind.
Given
- mw = 6.0 kg
- mc = 0.5 kg (8% ratio, mid-range)
- rp = 0.040 m
- g = 9.81 m/s²
Solution
Step 1 — at the nominal counterweight ratio of 8% (mc = 0.5 kg), compute the net driving force on the loop:
Step 2 — multiply by pulley groove radius to get the nominal driving torque at the great wheel:
That is the sweet spot for an Arnold-class regulator — enough headroom to drive the going train, the dial work, and the seconds beat through pivot friction without crowding the escapement.
Step 3 — at the low end of typical practice, a 3% counterweight (0.18 kg) you might see on a light Vienna regulator scaled down:
Only 6% more torque than nominal, but the loop runs slack on the counterweight side and the cord will jump the idler groove within a week. Pendulum amplitude looks fine until the cord derails — then it stops dead.
Step 4 — at the high end, a 15% counterweight (0.9 kg):
You have lost 7% of your driving torque to the counterweight. On a tight regulator with a heavy mercury-jar pendulum, that 7% can drop amplitude from 3° to roughly 2.4° and start to introduce circular-error rate variation. Stay near 8% and you keep the loop taut without bleeding torque.
Result
Nominal driving torque at the great wheel is 2. 16 N·m with a 6 kg main weight and 0.5 kg counterweight on a 40 mm-radius pulley. That's enough to swing a 1-second mercury pendulum cleanly at 3° amplitude with roughly 30% torque margin over the escapement's stall point. The full operating range from 3% to 15% counterweight only varies torque between 2.00 and 2.28 N·m — about 14% spread — but the sweet spot sits firmly around 8% because below that the loop derails and above it amplitude suffers. If you measure clock amplitude below 2.5° after fitting the new weight system, check three things in this order: (1) idler-pulley pivot friction — a sticky idler eats torque before it reaches the great wheel, (2) cord diameter mismatch to groove width which causes the cord to ride high and slip under load, and (3) click-spring tension set too stiff, which loads the ratchet teeth with parasitic friction during the running cycle.
Choosing the Endless Cord-winding Device for Clocks: Pros and Cons
The endless loop competes with two other ways of solving the winding-interruption problem: a going barrel (mainspring drives the train, winding tensions the spring without disconnecting drive), and Harrison-style maintaining power (a small auxiliary spring takes over for the 30 seconds of winding). Each has its place. The choice comes down to weight class, accuracy target, and how much you trust the user to wind on schedule.
| Property | Endless cord/chain loop | Going barrel (mainspring) | Harrison maintaining power |
|---|---|---|---|
| Continuous drive during winding | Yes — uninterrupted | Yes — uninterrupted | Yes, for ~30 seconds only |
| Typical driving load | 4 to 100+ kg weight | Spring torque equivalent to 0.5 to 3 kg | Whatever the host weight provides |
| Rate stability | Excellent — constant weight = constant torque | Variable — spring torque drops 30 to 50% across mainspring | Excellent (matches host system) |
| Maintenance interval | Cord renewal every 10 to 30 years | Mainspring service every 5 to 10 years | Auxiliary spring rarely fails — 50+ years |
| Mechanical complexity | Low — pulleys, ratchet, click, two weights | Medium — barrel, arbor, click, mainspring | Medium — added gear and spring on great wheel |
| Cost to build | Low — mostly turned brass and standard cord | Medium — requires precision spring and barrel | Low to medium — added to existing weight drive |
| Best application fit | Longcase, turret, regulator clocks | Spring-driven mantel, carriage, table clocks | Weight-driven precision regulators |
Frequently Asked Questions About Endless Cord-winding Device for Clocks
This is almost always idler-pulley friction increasing as accumulated dust and old gut lubricant migrate into the pivot. Constant weight should mean constant torque, so any amplitude drift over a winding cycle points to friction somewhere downstream of the weight — and the idler pivot is the most common culprit because nobody thinks to service it.
Lift the loop off, spin the idler by hand, and time how long it freewheels from a flick. A clean idler should run 5 to 10 seconds. Under 2 seconds means the pivot needs cleaning and a drop of clock oil.
Start at 8% of the main weight and verify two things: the loop sits firmly in both pulley grooves with no visible sag, and the cord doesn't slip when you push the great wheel backwards by hand. If it sags, go up in 1% steps until it tensions. If it slips, your groove profile is wrong, not your counterweight.
Above 15% you start bleeding driving torque you can't afford on a precision regulator. Below 3% the loop derails the moment a draught moves the case.
If the original movement was built with an endless loop, restore the endless loop — period accuracy matters and the loop is original to most pre-1750 longcases. If the movement is post-1760 English and the great wheel has the characteristic ratchet ring with a planted stud for an auxiliary spring, it was designed for Harrison maintaining power and that's what should be there.
For a new build, Harrison maintaining power is mechanically tidier and lets you use a normal barrel-and-line setup. The endless loop wins when you need to run weights above 20 kg or when winding intervals are weekly rather than daily — sustained continuous drive beats a 30-second auxiliary spring.
The click bounced off a ratchet tooth under shock and the loop unspooled freely. This usually means the click spring is too weak, the ratchet teeth are worn at the leading edge, or someone wound the clock too aggressively and set up a vibration that lifted the click.
Inspect the ratchet teeth under magnification — if the leading face is rounded rather than sharp, the wheel needs re-cutting. If the teeth are crisp, replace or re-tension the click spring to give roughly 0.3 N seating force at the click tip.
A knot is roughly 2 to 3× the cord diameter. When it reaches the pulley groove it cannot seat, so it lifts the cord clear of the groove for about 30° of pulley rotation. During those 30°, the click and ratchet may briefly disengage as the cord skips, and the great wheel loses the friction grip the loop normally provides.
The fix is a proper splice — taper-spliced gut or back-spliced polyester — never a knot. On chain loops, use a riveted master link, never a quick link.
Yes for most domestic work, with one caveat: polyester stretches less than gut under load (around 1% versus 3 to 5%) but it's also slipperier in the pulley groove. If your great-wheel pulley has a smooth half-round groove cut for gut, polyester may slip under shock loading. Re-cut the groove with a slightly sharper V-profile, or score the groove bottom with a fine triangular file to give the polyester something to grip.
Rate is unaffected because the loop only delivers torque — it doesn't time anything. The pendulum and escapement set the rate.
References & Further Reading
- Wikipedia contributors. Maintaining power. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
