Updated October 2026
Rate tells you if a watch runs fast or slow overall. Beat error measures something different and more subtle: whether the watch's tick-tock rhythm is evenly spaced, independent of whether the overall rate is accurate.
A healthy escapement produces alternating tick and tock sounds that are evenly spaced in time — the gap from tick-to-tock should equal the gap from tock-to-tick. Beat error is the difference between those two alternating gaps, expressed in milliseconds. This tool calculates it exactly that way: it splits the sequence of tick-to-tick intervals into two interleaved groups (the "even" gaps and the "odd" gaps, preserving their original alternating order) and takes the absolute difference between the two groups' averages — see how this works for the full mechanics, including why at least 3 clean intervals are required in each group before a beat-error number is shown at all.
Beat error originates in the escapement's geometry — specifically, how precisely the hairspring's collet (the small clamp that attaches the spring's inner end to the balance staff) and stud (where the outer end is fixed) are positioned relative to each other. When they're perfectly aligned, the balance wheel's rest position sits exactly centered between its two swing extremes, producing equal tick and tock gaps. When they're off-center — from manufacturing tolerance, wear, a past shock, or a prior repair — the balance rests slightly off-center, and one gap becomes shorter than the other even though the watch may still be running at a perfectly accurate overall rate.
This is the detail that trips people up: a watch can have excellent rate (very close to 0 seconds/day) and still have high beat error, or vice versa. Rate is about the average speed of the escapement over many beats; beat error is about the symmetry of each individual pair of beats. A watchmaker regulating a watch typically addresses both, but they're adjusted differently — rate via the regulator (changing effective hairspring length), beat error via repositioning the hairspring's collet.
The results panel flags beat error into three bands: under 1ms is "Excellent," 1-3ms is "Acceptable," and beyond 3ms is "High — may need service." These track reasonably well with general watchmaking convention, where well under 1ms is considered excellent regulation, a few milliseconds is tolerable for everyday wear without being alarming, and high single-digit or double-digit millisecond readings typically indicate the escapement genuinely needs attention.
Beat error itself doesn't necessarily ruin accuracy outright, but it tends to make a watch's rate less consistent and more position-sensitive than it would otherwise be — a watch with high beat error often shows a wider spread of rate readings across different positions (see testing positions) than the same watch would after the beat error is corrected. It's also sometimes an early indicator of a shock or impact the watch has taken, worth knowing about even if the rate itself still looks acceptable.
No — correcting beat error means physically repositioning the hairspring's collet on the balance staff, a precision adjustment done with specialized tools under magnification. This is watchmaker work, not something to attempt without training; this tool's job stops at measuring and flagging the number, not adjusting it.
Can beat error change over time on its own? It can shift after a shock, a drop, or general wear to the hairspring stud/collet assembly, though it's generally more stable day-to-day than rate is.
Why does beat error sometimes not show up in the results? The calculation needs at least 3 clean intervals in each of the two alternating groups; a short or noisy capture that doesn't produce enough reliable ticks in both groups will show rate without a beat-error number.
Is beat error the same thing professional timegrapher machines show? The concept and units (milliseconds) are the same industry-standard measurement; this tool derives it from open-air microphone audio rather than a calibrated clamped sensor, so treat it as a useful estimate rather than a certified figure.