A watch train is the chain of toothed wheels and pinions that transmits power from the mainspring barrel to the escape wheel in a mechanical watch. Watchmakers and movement designers depend on it to step a slow, high-torque input down into a fast, low-torque output that drives the balance wheel through the escapement. Each wheel-pinion pair multiplies speed and reduces torque by a calculated ratio, so the barrel turning once every few hours can drive a balance oscillating at 4 Hz. The result is a wristwatch keeping seconds-per-day accuracy from a single wound spring.
Watch Train Interactive Calculator
Vary the four wheel-pinion ratios and barrel period to see the total speed multiplication and resulting wheel speeds.
Equation Used
The total watch-train speed multiplication is the product of the four wheel-pinion ratios. Barrel speed is 1/T in rev/hr, and escape-wheel speed is that barrel speed multiplied by the total ratio.
- Ratios are wheel teeth divided by driven pinion leaves.
- Ideal gear mesh with no slip, backlash loss, or friction loss.
- Output speed is escape-wheel rpm, not balance vibrations per hour.
How the Watch Train Actually Works
The barrel inside a wristwatch turns slowly — typically one revolution every 6 to 8 hours when fully wound. The balance wheel needs to oscillate at 18,000, 21,600, 28,800 or 36,000 vibrations per hour depending on the calibre. Bridging that gap is the going train: a series of wheels and pinions in mesh, each pair acting as a small step-up gearbox. Power enters at the barrel, leaves at the escape wheel, and along the way one wheel turns once per hour (the centre wheel) and another turns once per minute (the fourth wheel, which carries the seconds hand on most movements).
The geometry is fixed by tooth counts. A typical ETA 2824-2 going train uses a centre wheel with 80 teeth driving a third-wheel pinion of 10 leaves, then a third wheel of 75 teeth driving a fourth-wheel pinion of 10 leaves, then a fourth wheel of 70 teeth driving an escape pinion of 7 leaves. Multiply those ratios and you land on the speed multiplication needed to feed a 28,800 vph balance. If even one tooth count is wrong the watch will not run at the correct rate — there is no calibration screw that fixes a miscut wheel.
Tolerances on pinion leaves matter as much as ratios. A leaf cut 0.01 mm undersize causes the wheel to ride too deep, increasing friction and dropping amplitude. Cut it 0.01 mm oversize and the teeth jam at the recoil. The most common failure modes you see on a service bench are pivot wear at the third- and fourth-wheel jewels, magnetised wheels disturbing the balance, and dried oil at the escape pinion causing amplitude loss before the watch outright stops.
Key Components
- Mainspring Barrel: Stores the energy and acts as the input wheel of the train. A typical wristwatch barrel carries 60 to 80 teeth and turns roughly once every 6 to 8 hours under load. Barrel arbor end-shake is held to about 0.02 to 0.04 mm — too tight and the spring binds, too loose and the barrel wobbles into the centre pinion.
- Centre Wheel: The first wheel after the barrel and the one that turns exactly once per hour, carrying the minute hand on most calibres. Tooth count is usually 64 to 80, meshing with a 10- or 12-leaf pinion. The centre pinion is the highest-torque pinion in the train after the barrel and is the most common location for tooth wear on watches that have run unserviced for 10+ years.
- Third Wheel: An intermediate step-up wheel with no timekeeping role. Typical tooth counts are 60 to 80 with a pinion of 8 to 10 leaves. Its job is purely ratio — it sits between the centre wheel and the fourth wheel and is where most of the speed multiplication happens.
- Fourth Wheel: Turns once per minute and drives the seconds hand on a centre-seconds or sub-seconds layout. Tooth counts run 60 to 80 with a 6- to 10-leaf pinion. Pivot diameter is typically 0.08 to 0.12 mm — these are the smallest jewelled pivots in the train and the first to show wear if the watch is run dry.
- Escape Wheel: The output wheel of the train and the input to the escapement. Almost universally 15 teeth on a Swiss lever escapement. The pinion is 6 to 8 leaves. Escape-wheel tooth geometry is the most precision-critical surface in the entire watch — impulse-face angles are held to within 0.5° of nominal.
- Pinion Leaves: The small driving gears on each wheel arbor, with 6 to 12 leaves. Leaf form is usually a modified ogive profile to reduce sliding friction at the mesh. A worn or miscut pinion drops amplitude by 20 to 40° before any other symptom appears.
Industries That Rely on the Watch Train
Every mechanical watch and most mechanical clocks contain a watch train of some form. The specific tooth counts change with the beat rate and the power-reserve target, but the architecture is the same from a £200 Seiko to a £200,000 Patek Philippe. You see the design problem most clearly in three places: high-frequency chronographs where the train has to feed a 36,000 vph balance, long-power-reserve calibres where the barrel turns even more slowly, and tourbillons where the escape wheel ends up on a rotating cage and the going train has to deliver power through the cage pinion.
- Wristwatch Movements: ETA 2824-2 and Sellita SW200 calibres use a 4-wheel going train feeding a 28,800 vph balance, found in millions of Swiss-made three-hand watches from Hamilton, Tissot, and Oris.
- High-Beat Watchmaking: Grand Seiko Hi-Beat 9S85 runs a 36,000 vph train requiring tighter pivot tolerances and a stronger mainspring to maintain 55 hours of reserve.
- Long Power Reserve: Lange & Söhne Lange 31 uses a twin-barrel going train delivering 31 days of running, with a constant-force remontoir between the train and the escapement.
- Chronograph Modules: Valjoux 7750 adds a separate chronograph train coupled to the fourth wheel via a vertical or lateral clutch — the going train must supply enough torque margin to drive the chrono without amplitude collapse.
- Tourbillon Cages: Breguet and Greubel Forsey tourbillons route power through a fixed fourth-wheel pinion into a rotating cage carrying the escape wheel and balance, demanding extreme concentricity at the cage bearings.
- Marine Chronometers: Hamilton Model 21 and Mercer chronometers use a fusee-and-chain ahead of the going train to flatten torque variation across the 56-hour reserve.
The Formula Behind the Watch Train
The total speed multiplication of a going train is the product of every wheel-tooth-to-pinion-leaf ratio in the chain. The number that matters at the design stage is the ratio between barrel rotation and escape-wheel rotation, because that ratio — combined with the escape wheel tooth count and the lock-and-impulse geometry — determines the balance frequency. At the low end of the typical range, a 4 Hz (28,800 vph) movement, the train multiplies barrel speed by roughly 3,600. At the high end, a 5 Hz (36,000 vph) Hi-Beat, the multiplication climbs near 4,500. Push beyond that and you start running out of mainspring torque to drive the train without amplitude collapse — that is the design ceiling.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Rtotal | Total speed multiplication from barrel to escape wheel | dimensionless | dimensionless |
| Zn | Tooth count of wheel n (barrel, centre, third, fourth) | teeth | teeth |
| Pn | Leaf count of pinion n (centre, third, fourth, escape) | leaves | leaves |
| fbal | Balance frequency derived from Rtotal × barrel speed × escape-tooth count / 2 | Hz | vph |
Worked Example: Watch Train in a 28,800 vph dress wristwatch calibre
A small movement-design studio in Glashütte is laying out the going train for a new in-house 28 mm dress calibre. Target balance frequency is 28,800 vph (4 Hz), target power reserve is 60 hours, and the barrel is sized to turn once every 7.5 hours at full wind. The escape wheel has 15 teeth as standard for a Swiss lever escapement. They need to verify that the candidate tooth counts deliver the correct balance frequency before cutting prototype wheels.
Given
- Zcentre = 80 teeth
- Pthird = 10 leaves
- Zthird = 75 teeth
- Pfourth = 10 leaves
- Zfourth = 70 teeth
- Pesc = 7 leaves
- Centre wheel speed = 1 rev/hour
- Escape teeth = 15 teeth
Solution
Step 1 — at the nominal design point, compute the ratio from centre wheel to escape wheel by multiplying the three wheel-to-pinion ratios:
Step 2 — the centre wheel turns once per hour (3600 s), so the escape wheel turns:
Step 3 — every escape-wheel revolution releases 15 teeth, and each tooth release corresponds to two balance vibrations (one per pallet stone), so:
That confirms the candidate tooth counts for the nominal 4 Hz target. Now check the operating range. At the low end of practical wristwatch beat rates — 18,000 vph (2.5 Hz) as you find in vintage calibres like the Omega 30T2 — the same architecture would need either a smaller centre-to-escape ratio or fewer escape teeth, dropping Rc→esc to about 70.3. At the high end, 36,000 vph as in the Grand Seiko 9S85, the ratio climbs to 140.6 and you must either add teeth to the centre and third wheels or run a stronger mainspring to feed the higher mesh losses. The 4 Hz layout shown here sits in the design sweet spot: enough rate stability to suppress positional error, low enough mesh velocity that pinion oil holds for a 5-year service interval.
Result
The candidate tooth counts deliver 28,800 vph exactly, with a centre-to-escape multiplication of 112. 5. In practice that means the escape wheel turns about once every 32 seconds and the seconds hand on the fourth wheel sweeps with 8 distinct ticks per second — visually smooth, audibly a clean 4 Hz tick on a timing machine. Across the operating range, an 18,000 vph version of the same architecture would feel mechanically lazy and lose amplitude faster off-wind, while a 36,000 vph version demands a 20-30% stronger mainspring and tighter jewel tolerances to avoid amplitude collapse below 250° at 24 hours. If your prototype measures 28,200 vph instead of 28,800, suspect three things first: a centre-wheel tooth count off by one (cut error), an escape wheel with 14 teeth instead of 15 (a surprisingly common stock-supplier mistake), or pinion-leaf form ground out of profile causing slip at the third-wheel mesh.
When to Use a Watch Train and When Not To
The watch train is one of several ways to step torque down and speed up in a portable timekeeper. The two real alternatives at the architectural level are a fusee-chain arrangement ahead of a conventional train, and a constant-force remontoir inserted between the train and the escapement. Each tackles a different problem.
| Property | Standard Watch Train | Fusee-and-Chain Train | Remontoir-Equipped Train |
|---|---|---|---|
| Power reserve typical range | 38-80 hours | 30-56 hours | 30-31 days (Lange 31) or 8 days typical |
| Rate stability across reserve | ±15 s/day drift between full and empty | ±3 s/day drift | ±1 s/day drift |
| Component count | 4-5 wheels | 4 wheels + fusee + 600+ chain links | 4 wheels + remontoir spring + secondary escapement |
| Movement thickness | 3.0-4.5 mm | 6-8 mm | 4.5-6 mm |
| Manufacturing cost (relative) | 1× | 8-15× | 5-20× |
| Service interval | 4-6 years | 3-5 years (chain wear) | 5-7 years |
| Application fit | Volume Swiss and Japanese calibres | Marine chronometers, ultra-high-end watches | Long-reserve and constant-force chronometry |
Frequently Asked Questions About Watch Train
Tooth counts set the average rate, not the daily rate. If the math is right and the watch still runs fast, the cause is almost always upstream of the train — the balance and hairspring. A hairspring sitting too short on the regulator stud, a balance with timing screws backed out unevenly, or magnetism in the spring tightening the active coil will all add 20-60 s/day without touching the going train.
Demagnetise first, then check the regulator pin gap — it should grip the spring with about 0.02 mm clearance. If you still see fast rate, weigh-check the balance against the design mass. A balance 1 mg light runs roughly 30 s/day fast on a 4 Hz movement.
Always add teeth where mesh velocity is lowest — that means the centre wheel. The centre-third mesh runs slower than the third-fourth or fourth-escape mesh, so a tooth-count change there adds the least friction. Adding teeth to the third wheel raises mesh velocity at two pinions (incoming and outgoing) and you lose more amplitude.
Rule of thumb: every 10% increase in third-wheel tooth count costs about 5° of balance amplitude at 24 hours. The same 10% on the centre wheel costs about 2°.
This is almost always a mainspring problem, not a train problem, but the train shows it first. A spring that has set or whose lubrication has dried delivers torque that falls steeply between full wind and the second day, even though the spring still uncoils for the rated reserve. The going train passes that torque drop straight to the balance.
Pull the barrel and measure free-state spring length. If it's more than 5% shorter than spec, replace it. If length is correct, regrease with Kluber P125 on the barrel wall — dry barrel walls cause stick-slip that mimics torque collapse.
You can in principle, and a few historical calibres did, but the geometry of a Swiss lever escapement is optimised around 15 teeth. Drop to 14 and the impulse angle changes, the lock depth shifts, and the safety roller clearances no longer line up with stock pallet forks. You end up redesigning the entire escapement, not just the wheel.
Unless you have a specific reason — Daniels-style independent double-wheel escapements, for example — stay at 15. The cost of changing every downstream tolerance outweighs any benefit from the gear-ratio flexibility.
Positional error of more than 10 s/day between horizontal and vertical points to pivot or jewel issues in the going train, not the balance. When the watch sits vertically, the wheel arbors load the side of their jewels rather than the endstone, and any out-of-round pivot or worn jewel hole adds friction unevenly through the rotation.
Inspect the third- and fourth-wheel pivots under 40× magnification. A pivot worn into a slight oval will look fine end-on but will catch under side load. Replacement is the only fix — burnishing only works on minor wear.
Centre distance between wheel and pinion is typically held to within ±0.01 mm of nominal. Beyond that and you see one of two failures: too close and the teeth bind at the deepest point of mesh, dropping amplitude by 30-50°; too far and the teeth skip under recoil, which sounds like a faint rattle on a microphone-equipped timing machine.
Use a depthing tool to set centre distance during plate manufacture, then verify with a feeler test — the wheel should turn the pinion with no perceptible backlash but no drag. If you can feel either, redo the jewel hole position.
References & Further Reading
- Wikipedia contributors. Watch. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
