Escapements: the hardest problem in watchmaking

An escapement has two jobs and they get in each other's way. It must hold back a gear train that is permanently trying to spin, releasing it one tooth at a time so that time is measured in equal pieces. It must also give the oscillator a push on every swing, because a balance wheel left alone would slow and stop within a minute. Locking wants a firm, secure engagement. Impulsing wants a light, frictionless touch. Every escapement ever built is a compromise between those two demands.

Add a third constraint and the problem becomes properly hard. Energy spent in the escapement is energy not delivered as running time, friction needs lubrication and lubrication ages, and any contact with the balance while it swings freely disturbs the oscillation you are trying to measure. The escapement is described in outline in our guide to how a mechanical watch works; this is the longer answer.

The verge: three centuries of not very good

The verge escapement appeared in European tower clocks around the end of the 13th century and is the ancestor of everything that follows. A vertical staff, the verge, carries two flags set at roughly a right angle, and a crown wheel with saw-like teeth pushes alternately on each, rocking the verge back and forth.

Its defining flaw is that an escape wheel tooth is in contact with a pallet for essentially the whole swing. This is a frictional rest escapement: the oscillator never runs free. Worse, the escape wheel is driven briefly backwards at the end of each swing, which is called recoil, so the whole train reverses direction thousands of times an hour.

The consequences are predictable. The rate depends heavily on how hard the mainspring is pulling, so accuracy varies through the day, and wear is fast. A good verge pocket watch of the 18th century might be out by several minutes a day. What kept it in production well into the 19th century is that it is cheap, forgiving of poor tolerances and almost impossible to stop.

The clock detour: anchor and deadbeat

Pendulum clocks took a different path, and the vocabulary is worth borrowing. The anchor escapement, in use from the 1660s and associated with Robert Hooke and the clockmaker William Clement, let a pendulum swing in a narrow arc while a rocking anchor caught the escape wheel teeth. It still recoiled.

George Graham's deadbeat escapement, from around 1715, removed the recoil by giving the pallets a locking face concentric with the pallet arbor. A tooth resting on that face exerts no turning force on the pendulum at all: the wheel stops dead, hence the name. That is the principle the rest of this story circles around. What matters is not eliminating contact, but making contact harmless.

The cylinder: elegant, small, and doomed to wear

The cylinder escapement, patented in outline by Thomas Tompion and colleagues in 1695 and made practical by Graham in the 1720s, replaced the verge in better watches for over a century. The balance staff carries a hollow steel cylinder with a section cut away. Escape wheel teeth enter the cylinder, are held by its wall, and are released as the cylinder rotates, pushing on the cut edge as they go.

It is beautifully compact, which is why it made flat pocket watches possible, and it does not recoil. But it is still a frictional rest design, and the whole load of the train bears on a thin cylinder wall that is also the balance's arbor. Steel cylinders wore into visible grooves; ruby cylinders helped and cost a great deal. It is horology's clearest case of a design that solved a shape problem and lost on a materials problem.

The detent: the most accurate escapement nobody wears

The detent escapement, also called the chronometer escapement, was developed by Pierre Le Roy in the mid-18th century and brought to its practical form by John Arnold and Thomas Earnshaw in the 1780s. It is the first genuinely detached escapement: for almost the entire swing, the balance is connected to nothing.

The mechanism is spare. A slender spring, the detent, holds the escape wheel locked. As the balance passes through center in one direction, a jewel on its staff pushes the detent aside, one tooth escapes and impulses a second jewel on the balance staff directly, and the detent springs back to lock the next tooth. On the return swing, a hinged passing spring lets the balance jewel brush past without unlocking anything.

Impulse is therefore delivered once per complete oscillation rather than twice, directly, radially, with almost no sliding, and the balance is otherwise free. It is generally regarded as the most accurate mechanical escapement ever put to practical use, and it dominated marine chronometers for a century and a half.

Two flaws kept it off wrists. It does not self-start: stop the watch and the balance sits there until it is physically shaken into motion. And under a sharp knock the detent can be displaced at the wrong moment, letting a tooth escape without an impulse or several escape at once, which the trade calls tripping. A marine chronometer lives in a gimballed box and is never dropped. A wristwatch is dropped roughly once a decade.

The Swiss lever: the design that won

The lever escapement, invented by Thomas Mudge around 1754 and refined into its Swiss club-tooth form over the following century, is what is inside almost every mechanical watch made today. It is detached, self-starting and safe, and those three properties together beat everything else on the list.

Here is the cycle. The pallet fork sits between two banking pins that limit its travel, with one of its two jeweled pallet stones locked against a tooth of the escape wheel, so the train is frozen. The balance swings back toward center and a jeweled impulse pin on its roller enters the fork's notch, knocking it across. The escape wheel is released, its tooth slides along the sloped impulse face of the pallet stone, and through the fork that push passes to the impulse pin and into the balance. The fork reaches the opposite banking pin, the other stone catches the next tooth, and the train locks again. The gap between unlocking and the next lock is drop, which is wasted energy, so designers minimize it without making tolerances impossible.

The balance then carries on alone, swinging out perhaps 300 degrees and back, with the fork still and touching nothing. That detachment is where the accuracy comes from.

Why draw is the clever part

The safety of the lever escapement rests on a feature that is easy to overlook. The locking faces of the pallet stones are not radial: they are angled slightly, so that the escape wheel's pressure on that angled face actively pulls the fork harder against its banking pin. This is draw, and it is why a lever watch survives being knocked. Without it, a jolt could throw the fork across and release the train while the balance is nowhere near. With it, any displacement is fighting the mainspring itself.

Draw is backed by two more safety features. The guard pin, a fine pin on the fork's tail, rides just clear of a circular safety roller on the balance staff, so a fork that tries to move at the wrong moment is stopped dead. A crescent-shaped notch in that roller lines up with the guard pin only at the instant the impulse pin is entering the fork, which is precisely when the fork is supposed to move.

The price for all this is sliding friction on the impulse faces, which needs oil, and oil that degrades is the main reason a watch needs servicing every few years.

Modern alternatives

Two lines of attack have reached series production in the last thirty years. One changes the geometry; the other changes the material.

The co-axial escapement

The English watchmaker George Daniels worked from the observation that the lever's weak point is sliding friction during impulse, and that this friction is what ties lever performance to lubricant behavior. His answer, patented in 1980 and put into series production by Omega from 1999 in the caliber 2500, is the co-axial escapement.

It uses an escape wheel with teeth on two levels, mounted co-axially, and a pallet fork carrying three stones rather than two. Two stones do the locking, as usual. Impulse is separate: in one direction a tooth of the upper wheel acts on the third pallet stone, and in the other it acts directly on an impulse roller on the balance staff.

The change is geometric. A lever escapement impulses along a long sloped face with substantial sliding. The co-axial impulses with a much more radial push over a shorter contact, closer to what a detent does, while keeping the self-starting, guarded structure of a lever. Less sliding means less dependence on oil at the impulse surfaces, which is the claimed benefit for service intervals.

It is not free: the parts count is higher, tolerances are tighter, adjustment needs specific training, and the escapement is fussier about the torque it receives.

Silicon: solving friction by changing the material

The other line of attack leaves the geometry alone and changes what the parts are made of. Silicon components, etched from wafers by deep reactive ion etching rather than cut and hardened, entered production watchmaking in the early 2000s: Ulysse Nardin's Freak of 2001 used silicium escapement wheels, and Swiss research and manufacturing collaborations brought silicon escape wheels, pallets and hairsprings into wider use over the following decade.

What silicon offers is real. It is not magnetic, which matters more than most people expect. It runs against a jewel with low enough friction to work without lubricant at the escapement. It can be etched to tolerances tighter than machining allows, in shapes the process does not charge extra for. And it is very light, which reduces escape wheel inertia.

What silicon costs is repairability. A silicon part cannot be adjusted, bent or reshaped at the bench: it is brittle, with no plastic deformation at all, so it either holds its shape or it fractures. Nothing about it can be corrected by a watchmaker's skill. When something goes wrong the assembly is replaced with a new one from the manufacturer, which makes servicing dependent on parts supply from a single source, possibly decades later. That is a genuine trade of one kind of longevity for another.

Constant force: the neighboring problem

Two devices are often discussed alongside escapements and are not escapements at all. They sit earlier in the chain, addressing the fact that a mainspring pulls harder when fully wound than when nearly run down.

A fusée is a cone-shaped, spirally grooved pulley connected to the barrel by a chain. As the spring weakens, the chain works on a progressively larger radius of the cone and the leverage compensates. It works well and takes up an enormous amount of space, which is why it survives only in a handful of very expensive watches.

A remontoire is a small secondary spring or weight between the train and the escapement, delivering a fixed impulse and being rewound by the mainspring at regular intervals, often once a second. The escapement never knows how wound the watch is.

Both improve isochronism, the property of a rate that does not change with amplitude. Neither changes what the escapement does. A marketing line about "constant force" describes this part of the movement, not the escapement itself.

Comparison

Why the lever keeps winning

There is a commercial reason and an engineering reason, and they point the same way.

The engineering reason is that the lever escapement is not actually the limiting factor in a modern watch. A well-made lever movement, properly adjusted, keeps time to within a few seconds a day, and that figure is set more by the balance, the hairspring and the positional adjustment than by the escapement geometry. Replacing it buys a smaller improvement than the effort suggests.

The commercial reason is that escapement innovation is nearly impossible to sell. It is invisible under a dial, it takes years and a great deal of money to industrialise, it requires retraining every watchmaker in the service network, and the benefit (a longer interval between services, a slightly steadier rate) is exactly the sort a customer cannot verify. A new dial color produces measurable sales next quarter. A new escapement produces a footnote.

That is not an argument that the co-axial and silicon work were wasted: both shifted what a mainstream movement can promise. It is an honest account of why a 200-year-old geometry, refined into something machines can make to a few microns, remains the escapement in almost every watch you will handle. As our history of the wristwatch shows, the industry changes what it can sell far faster than it changes what works.