How a mechanical watch works

A mechanical watch has one problem to solve: it holds a lump of stored energy that wants to release all at once, and it needs that energy to come out in equal, countable pieces instead. Everything inside the case exists to enforce that. Once you see the movement as an energy pipeline with a metronome bolted to the end of it, the vocabulary stops being intimidating.

There are five subsystems, and energy passes through them in order.

1. The mainspring: where the power lives

A mainspring is a long, flat ribbon of spring alloy, historically hardened steel and in most modern watches a cobalt-based alloy such as Nivaflex, coiled inside a toothed drum called the barrel. Winding the watch coils the spring tighter around the central arbor. The spring then spends the next day or two unwinding, turning the barrel, and the barrel's teeth drive the first wheel of the gear train.

Two numbers describe it. Power reserve is how long the watch runs from fully wound to stopped: roughly 38 to 42 hours on a great many movements, because that is what a single barrel of conventional proportions gives you. Modern designs push to 70 or 80 hours by using a longer, thinner spring in a wider barrel, by using two barrels, or by lowering the balance frequency so less energy is spent per hour. Torque is how hard the spring pulls, and it is not constant: a fully wound spring pulls harder than an almost-flat one.

That torque curve matters more than it sounds. Feed a balance wheel more power and it swings further; feed it less and it swings less, which changes the rate. This is why a watch that keeps beautiful time on day one can drift on the morning of day two, and why most manufacturers quote accuracy figures over the first 24 hours of a full wind.

Three classical answers exist to the falling-torque problem. A fusée and chain, a cone-shaped pulley that changes leverage as the spring runs down, solved it mechanically in the earliest spring-driven timepieces (the fusée goes back to the 15th century) and survives today only in a handful of very expensive watches. A stop-work simply limits how far the spring may be wound and unwound, keeping it away from its strongest and weakest turns. Most modern watches take the third route: accept the variation and make the escapement and balance tolerant of it.

2. The gear train: dividing the day down

The barrel turns slowly, a handful of turns per day. The balance wheel beats several times per second. Between them sits the going train, usually four wheels, each meshing a large wheel with a small pinion so that speed multiplies and torque falls at every step.

A conventional layout runs barrel, then center wheel (one turn per hour, which is where the minute hand often comes from), then third wheel, then fourth wheel (one turn per minute, which is where a small seconds hand comes from), then the escape wheel. Change the tooth counts and you move the seconds to the center or the minutes off-center, which is exactly what a movement designer does when the dial layout demands it.

Every one of those pivots turns in a jewel, a synthetic ruby with a polished hole in it. Jewels are not decoration and not precious: synthetic corundum has been cheap since the Verneuil process was industrialised early in the 20th century. They are there because a hard, smooth, oil-retaining bearing surface wears far less than steel on brass. A typical modern automatic carries 21 to 27 of them. A higher count is not automatically better, and a very high count on a simple movement usually means jewels have been added where they do nothing.

3. The escapement: releasing energy in countable pieces

This is the heart of the thing, and it is the part most descriptions skip.

The gear train is always trying to spin. The escape wheel at the end of it is held by a pallet fork, a small anchor-shaped lever with two jeweled pallet stones. The fork blocks the escape wheel, so the whole train freezes. Then the balance wheel, swinging back, knocks the fork out of the way. One tooth of the escape wheel slips past ("escapes", hence the name), the fork swings across, and the next tooth is caught. In the same motion, the escape wheel tooth pushes the fork, and the fork pushes the balance, giving it the tiny kick it needs to keep swinging.

So the escapement does two jobs at once: it locks the train so time is measured out in equal pieces, and it impulses the balance so the oscillation never dies. That double duty is why escapement design is the hardest problem in watchmaking.

The near-universal answer is the Swiss lever escapement, refined through the 18th and 19th centuries and dominant ever since. It is robust, it self-starts, it tolerates shocks, and it can be made by machine to tight tolerances. Its weakness is that impulse happens by sliding contact, so it needs oil, and the oil ages.

The best-known modern alternative is the co-axial escapement, designed by the English watchmaker George Daniels and put into series production by Omega from 1999. It separates locking from impulse across two levels of escape wheel so that impulse is delivered by a more radial push and less sliding, cutting friction at the contact surfaces. Silicon components, detent-style escapements in high-end pieces, and various proprietary geometries all chase the same goal from different directions.

4. The balance and hairspring: the metronome

The balance wheel is a weighted ring on a staff, and the hairspring (or balance spring) is a fine coiled spring attached to it. Together they are a torsional oscillator: displace the wheel, and the spring pulls it back, and it overshoots, and it comes back again. Left alone it would swing at a rate set almost entirely by the wheel's inertia and the spring's stiffness. That rate is what the watch counts.

Frequency is quoted in vibrations per hour (vph) or in hertz, and one hertz is two vibrations because each full swing has a tick and a tock.

Higher frequency means each error is a smaller fraction of a second and the balance recovers faster from a knock, so high-beat movements tend to be more stable on the wrist. The cost is wear and lubrication: everything happens more often, so the escapement works harder and service intervals can be shorter.

Amplitude, measured in degrees, is how far the balance swings from center. A healthy movement typically shows somewhere around 270 to 310 degrees dial-up when fully wound, dropping perhaps 40 to 50 degrees by the end of the power reserve and falling further in vertical positions. Amplitude is the single most useful health indicator a watchmaker reads off a timing machine, because low amplitude points at old oil, a tired mainspring or a fault long before the rate itself misbehaves.

Two refinements are worth knowing by name. A shock protection system (Incabloc and KIF are the classic trade names) mounts the delicate balance staff pivots in spring-loaded jewel settings so a knock is absorbed by the whole assembly moving rather than by the pivot snapping. And regulation is how the rate is trimmed: a traditional regulator arm shortens or lengthens the working length of the hairspring, while a free-sprung balance leaves the spring alone and adjusts inertia with weights or screws on the rim, which is more stable and more expensive to set.

5. The motion works and the dial train

The gear train's job ends at the escapement. Driving hands is a separate, slower set of wheels called the motion works, sitting under the dial. It takes the hour-rate output and gears it 12:1 so that the hour hand turns once for every twelve turns of the minute hand.

The clever detail is the cannon pinion, a friction-fitted tube on the center arbor. It is tight enough to be driven normally and loose enough to be twisted independently, which is precisely what happens when you set the time: the hands move while the going train underneath carries on. Without that friction fit you could not set a watch without stopping it.

6. The keyless works: winding and setting

Pull the crown out and something has to change what the crown does. That switching mechanism is the keyless works: a sliding pinion, a yoke, a setting lever and a detent spring, arranged so that crown position selects a function.

• Pushed in: the crown winds the mainspring through the winding pinion, crown wheel and ratchet wheel.
• First click out (on watches that have one): a quickset, usually for the date.
• Fully out: the sliding pinion engages the motion works and the crown sets the hands.

Many movements add a hacking lever that touches the balance rim when the crown is fully out, stopping the watch dead so you can set it to a time signal. If your seconds hand does not stop, the movement simply lacks that lever; it is not a fault.

Putting it together: one second of a watch's life

At 28,800 vph, in one second: the balance swings out and back four times, so the escapement unlocks eight times. Each unlocking lets the escape wheel advance by about half a tooth space, so four teeth pass in that second. The fourth wheel has moved one sixtieth of a revolution and the seconds hand has advanced one second division, which is why a mechanical seconds hand looks like it sweeps but under a loupe is visibly stepping eight times a second.

Repeat 86,400 times and you have a day. If the packets are on average a hair too big, you gain seconds; a hair too small, you lose them. Everything in the accuracy standards guide is about how small "a hair" is allowed to be.

What actually goes wrong

Understanding the chain tells you where a sick watch is sick.

A dropped watch that suddenly runs badly has usually had its balance staff or shock setting disturbed. A watch that suddenly runs fast has usually been near a speaker, a laptop, a magnetic clasp or a tablet cover, and demagnetising takes a watchmaker seconds.