The Escapement

Hold a mechanical watch to your ear and the first thing you notice is that it is not smooth. A spring is unwinding inside it, a steady continuous pull, and yet what you hear is a stutter — tick, tick, tick — force arriving in identical grains rather than a flow. That stutter is the whole secret of the thing. Somewhere between the wound spring and the moving hands sits a small mechanism whose entire job is to refuse the spring's smoothness: to stop the wheels almost all of the time, and let them advance one tooth, and stop them again. It is called the escapement, and the name is exact. The word came into English in the late 1700s from the French échappement, from échapper, to escape — which runs back through Latin to the image of a man slipping his pursuer's grasp by wriggling out of his own cloak, leaving the cloak behind in the hand that grabbed it. Each beat, one tooth of the escape wheel slips the grip of the pallets and gets away. The clock runs on a long series of small escapes.

The mechanism has two jobs and does both in the same instant. It is the gate between the going train — the geared wheels driven by the falling weight or the unwinding mainspring, supplying a continuous torque — and the oscillator, the pendulum or the balance wheel that actually keeps the time. Most of every beat the escapement holds the train locked, dead still, the whole driving force pressing against a single tooth that cannot move. Then, for a fraction of a second, it releases: one tooth escapes, the train lurches forward by exactly that much, and in the same motion the escaping tooth delivers a tiny push to the oscillator — just enough to replace what friction and air have stolen from the last swing, so the swinging never dies. Lock, release, impulse, lock. What opens the gate, each time, is the oscillator's own rhythm. The clock does not impose time on its motion; it counts a motion it does not create and must not disturb.

That last clause is the deep design constraint, and it is stricter than it sounds. The pendulum keeps good time because it is very nearly isochronous — its period barely changes whether it swings wide or narrow, whether the drive is strong or weak. But the escapement, in order to keep the pendulum alive, has to touch it, and every touch is a chance to corrupt the period it is supposed to be reading. So the whole art is to touch as little, as briefly, and as symmetrically as possible. The cleanest place to deliver the push is at the very bottom of the swing, the equilibrium point, where the restoring force is zero and the oscillator is moving fastest — a principle the astronomer George Airy made precise in the nineteenth century: impulse applied symmetrically at that zero-crossing leaves the period very nearly unchanged even as the amplitude wanders. The best escapements are the ones that intrude least. They leave the oscillator detached, swinging free through most of its arc, and slip in only for the instant of the push. Accuracy, here, is a kind of restraint — the discipline of a mechanism that exists to interfere and is judged by how little it does.

The history of clocks is the history of learning that restraint, and it reads as a catalogue of which flaw mattered most at each level of precision. The first escapement, the verge, appeared in Europe around 1300 and ruled alone for nearly four centuries; it let its bar swing through eighty or a hundred degrees and kicked the escape wheel backward a little on every beat — recoil — so it was forever fighting itself, and a good verge clock might lose many minutes a day. The anchor escapement, attributed to Robert Hooke around 1657 and spread by William Clement's longcase clocks by 1680, shaped its pallets like a ship's anchor and cut the pendulum's swing to four or six degrees — closing in on the small arc where a pendulum's near-isochronism actually holds, since a pendulum swung wide runs slower than one swung narrow — which let the long seconds-pendulum tick once a second and dragged the daily error down to minutes. Then came the deadbeat — pallet faces curved so the locked tooth rested truly dead, no recoil at all — perfected around 1715 by "Honest George Graham," who never patented his escapements, being more interested in moving science forward than in owning it, and who quietly lent money to a self-taught Lincolnshire carpenter named John Harrison.

Harrison was chasing the hardest version of the problem. A clock at sea, on a pitching deck, through tropical heat and arctic cold, was the only practical way to find a ship's longitude — the Board of Longitude had put up as much as twenty thousand pounds for a solution — and every error the land clocks had tamed came roaring back at once. Harrison answered them one at a time. His grasshopper escapement, developed in the 1720s, was nearly frictionless, its arms kicking like an insect's legs so that almost no power was lost and almost no rust could bind it. His gridiron pendulum cancelled the way metal lengthens in heat. By the time he shrank it all into the watch-sized H4, finished in 1759, he had a machine that on its 1761 trial to Jamaica — his son William carrying it aboard HMS Deptford, Harrison himself nearly seventy and too frail to go — lost about five seconds over eighty-one days. The Board suspected a fluke and made him prove it again. Around the same time Thomas Mudge worked out the detached lever escapement, which leaves the balance almost entirely free and which still ticks inside nearly every mechanical watch made since.

What every one of them was really building was a better way to stop. We picture a clock as a thing that runs, and we picture time as a thing that flows, and the escapement quietly contradicts both. The water clocks of the ancient world measured time by flow, and they were hopeless, because flow has no edges and nothing in it is the same as anything else. The mechanical escapement's whole revolution was to throw the smoothness away — to chop the continuous push into a long file of discrete, identical, countable units, and to make accuracy a matter of the sameness of the units rather than the steadiness of the push. The tick you hear is not, mostly, the sound of the clock releasing its force. The loudest part of it is the force being caught again — the escaped tooth running forward and slamming against the next pallet, the train seized back into stillness an instant after it was freed. The clock spends almost all of its life holding its power back, and the proof that it is working is that small recurring sound of arrest. It does not keep time by moving. It keeps time by the regularity of its stopping.