A recoil escapement pushes the train backwards on every beat. The cost is small per beat and large per year, and it grows when the drive force varies — which is the part that matters.
What recoil costs, worked per beat
issue ISSUE 01written 2026-07-24responsible Escapement Papers (a named bench, not a person)
tape card: recoil worked out
| Interval | Beats | Recoil cost, order of magnitude |
|---|---|---|
| One beat | 1 | A small fraction of a millisecond |
| One day, 1 s pendulum | 86,400 | Tens of seconds |
| One year | 31.5 million | Minutes to hours |
| Per year, drive force varying | 31.5 million | The same, plus the drift |
The arithmetic, and why it is a range
Recoil costs time because the escape wheel moves backwards before it moves forwards. The size of the loss depends on the pallet geometry and on how hard the train is pushing, so a precise figure has to be measured for a particular clock. What can be said without measuring is the scale: the loss is per beat, and a seconds pendulum has 86,400 beats in a day.
| Quantity | Value | What it does to the argument |
|---|---|---|
| Beats per day, seconds pendulum | 86,400 | Multiplies every per-beat loss |
| Typical recoil per beat | a fraction of a millisecond | Small in isolation |
| Daily effect | tens of seconds | Comparable to a whole rate specification |
| Effect of a varying drive | the same again | The part the deadbeat removes |
Why the varying part is the important part
A constant recoil can be regulated out. A recoil that changes as the driving weight descends cannot, because the regulator is set once and the force is not constant. That is why the deadbeat escapement mattered more than its raw accuracy figure suggests: it removed the dependence, not just the loss.
This is the same argument the lever escapement faces in a different form. The balance is in contact with the escapement during impulse, so the train's condition enters the rate. Detachment is the answer, and it is why chronometers used a detent.
A sentence to argue with
The bench holds that the recoil escapement's accuracy is understated in popular writing, because a recoil clock with a constant drive force beats a deadbeat clock with a varying one. The claim is contestable, and it is testable with two clocks and a note of the drive force over a week.
References
- Ref 1Main plate, Horopedia – Watchmaking Encyclopediasupports the main plate on which the recoiling train is mounted
- Ref 2Barrel Drum, Horopedia – Watchmaking Encyclopediasupports the barrel whose stored energy drives the recoil
- Ref 3Barrel Arbor, Horopedia – Watchmaking Encyclopediasupports the barrel arbor that transmits the drive force the recoil depends on
One link, one document, titled as it titles itself. No home pages, no search results.
Set this page down Set this page down when you know whether the clock's drive force is constant. That decides whether recoil matters.