Timeless TicksWatch knowledge, checked Search Menu

Glossary · also called beat, out of beat

Beat error

Definition

The difference in duration between a balance wheel's two half swings, measured in milliseconds, with anything under about 0.5 ms generally regarded as good.

On this page (6 sections)
  1. How a timing machine shows it
  2. What good looks like
  3. Why it matters
  4. Where it comes from
  5. In practice
  6. Common misconceptions

Beat error is the timing difference between a balance wheel's swing in one direction and its swing back in the other, expressed in milliseconds. In a perfectly set movement the tick and the tock are evenly spaced and the beat error is zero. When the balance's rest position is not aligned with the escapement's line of centers, one half oscillation takes slightly longer than the other, and a timing machine hears the resulting limp as a pair of traces separated by a measurable gap.

The alignment that matters is between the balance's rest position and the line through the centers of the balance, the pallet fork and the escape wheel. At 28,800 vibrations per hour each half swing lasts about 125 milliseconds, so 0.5 ms of error is an asymmetry of four parts in a thousand, which at a healthy 280 degrees of amplitude works out at roughly one degree of misalignment.

How a timing machine shows it

A timing machine listens through a microphone clamp and plots each impulse as a dot. Two traces appear, one for the tick and one for the tock, and the vertical gap between them is the beat error. The timegrapher reading explainer covers all three readings together.

Two things move the number without anything being wrong. Beat error is read against amplitude, so the same misalignment looks worse as the swing shrinks: 0.4 ms at full wind can read 0.6 ms near the end of the reserve. A slower balance also stretches every interval, so an identical offset reads about a third higher at 21,600 vph than at 28,800. Measure at full wind, in a noted position.

What good looks like

Under about 0.5 ms is generally regarded as good for a wristwatch, and a well-adjusted movement is often set below 0.3 ms. Figures above roughly 1 ms are worth correcting. Some tolerance is normal and, unlike rate, a modest beat error does not by itself make the watch run fast or slow.

Beat error at full windWhat it means
Under 0.3 msProperly set, and the state a movement should leave a bench in.
0.3 to 0.5 msGood. No practical consequence at normal amplitude.
0.5 to 1.0 msUsually acceptable. Worth correcting if the movement is open anyway.
Above 1.0 msCorrect it. Self-starting suffers first, normally at low wind.

These are workshop conventions, not a published standard. No chronometer test measures beat error: the seven COSC criteria are all rate criteria, so a movement can be certified with a beat error a watchmaker would tidy up on sight.

Why it matters

The practical consequence is starting and stopping rather than daily rate. A movement well out of beat receives uneven impulses, so it may fail to restart on its own after stopping, or stall at low amplitude near the end of the power reserve. Amplitude in the two directions can also become unequal, which makes the whole timing trace harder to interpret.

The failure is specific. Out of beat, a movement still runs well while wound and moving. The trouble comes at rest: set the watch down 30 hours into a 42 hour reserve and it can stop, then decline to restart when picked up. Owners meet this as a watch that stops overnight, which has several causes, beat error among them.

Where it comes from

Beat error does not creep up the way amplitude does. It is a position error, so it usually arrives in one event: a knock that shifts the hairspring collet on the staff, a balance refitted without checking beat afterwards, a replacement balance complete fitted from the packet, or a disturbed stud. A watch back from an unfamiliar workshop reading 1.5 ms was probably assembled that way rather than damaged since.

In practice

Correcting beat error means rotating the hairspring collet on the balance staff, or moving the stud carrier on movements built with an adjustable one. It is a delicate job on a fixed-stud movement, since the collet must be turned without distorting the spring, and it is usually done as part of a service rather than on its own.

A movable stud carrier, fitted to many modern calibers and to most watches with a free-sprung balance, lets the correction be made with the balance in place while the machine watches. A fixed stud means working largely by feel, which is where hairsprings get distorted.

"Please regulate it" buys rate adjustment only, so say you want the beat error corrected and the watch timed in several positions. Correction is usually folded into a full service, an indicative $250 to $500 for a three-hand automatic at an independent watchmaker in the United States before sales tax, moving with market and region: see what a watch service costs.

Common misconceptions

Beat error is not the same as rate, and a watch with a perfect zero can still run ten seconds a day fast. Nor is it the same as amplitude, though the two are read from the same machine and a large beat error will often be found alongside other symptoms of a movement needing attention.

The bigger trap is the word beat. A movement beating at 28,800 beats per hour is being described by its frequency, not its beat error: see vibrations per hour and the beat rate converter. Zero is not worth chasing either, since 0.0 ms on a hobby machine usually means the display has rounded. Our guide to timing your own watch covers what to measure before deciding anything needs doing.

Last reviewed 4 September 2026. Spotted an error? Tell us and we will fix it in public.