The torque variation problem
Set a fine watch on a timing machine, measure it fully wound, leave it on the bench for forty hours, and measure again. The difference between the two rates — usually a few seconds a day, sometimes more — is the signature of a problem watchmakers have been attacking since the very first spring-driven clocks. A fully wound mainspring delivers more torque to the gear train than a half-wound one; that is simply what a spring is. As it runs down over its reserve, the force reaching the escapement falls continuously, the balance's amplitude falls with it, and the rate drifts — because no balance is perfectly isochronous across amplitudes.
For a simple three-hand watch the variation is manageable, and most wearers never notice. For precision timekeepers it is a fundamental limit. And for complications that depend on consistent force — striking mechanisms above all, whose hammers must land with reliable energy — it can mean malfunction at low reserve. The history of constant force is the history of trying to make a spring behave like a battery: same output, first hour to last.
The simple answers: stop-works and the going barrel
Not every solution is complex. The Geneva stop-work — a Maltese-cross-and-finger mechanism on the barrel — simply forbids the mainspring from operating outside the middle of its travel, excluding both the harsh torque of full wind and the feeble torque of near-exhaustion. It appears in fine watchmaking from the eighteenth century into the twentieth (Patek Philippe's early Calatrava calibres carried one) and survives today mostly in movements that take their chronometry ceremonially seriously. Its sixteenth-century German cousin, the stackfreed — a cam-and-spring brake that resisted the spring hardest at full wind — was cruder and died out quickly. The modern mainstream answer is the opposite of clever: the going barrel, no stop-work at all, paired with a long mainspring used in its flattest middle region, a well-designed escapement, and a balance adjusted to tolerate amplitude change. Most quality modern movements simply engineer the problem small rather than solving it outright.
The fusee chain
The aristocratic solution is as old as the portable spring itself. The fusee — a cone-shaped, spirally grooved pulley connected to the barrel by a miniature chain — appears in the earliest surviving spring clocks of the mid-fifteenth century and remained the backbone of English precision watchmaking into the twentieth. The geometry is the whole idea: at full wind, the chain pulls on the fusee's narrow end, where leverage is least; as the spring weakens, the chain works down to the wide end, where leverage is greatest. Falling force, rising mechanical advantage — calculated to cancel, delivering nearly constant torque across the reserve, against the 30 to 50 percent variation of a bare barrel. The chain itself is a marvel of pre-industrial production: hundreds of links and rivets, hand-assembled, the whole assembly narrower than a millimetre yet carrying the mainspring's full pull. Add Harrison's maintaining power — a small auxiliary spring keeping the train driving while the fusee is being wound — and you have the package that powered the marine chronometer era.
Its costs are bulk and complexity, which is why it vanished once wristwatch thinness became commercially decisive. Its revival in modern wristwatches — A. Lange & Söhne's Richard Lange "Pour le Mérite" series, Romain Gauthier's Logical One (which substitutes a ruby-linked chain and snail cam) — is part engineering achievement and part horological argument: a statement that rate consistency deserves the space it costs.
The remontoire
The remontoire takes the opposite approach: instead of taming the mainspring, it isolates the escapement from it. A small secondary spring sits between the train and the escapement and is rewound at fixed intervals — every second, or every minute — by the mainspring's energy. The escapement is driven only by this little spring, which is always at essentially the same tension when it delivers, so the balance receives identical impulses whether the mainspring is fresh or nearly flat. The cost is real complexity: the remontoire spring, its tripping and rewinding mechanism, and the governance that keeps everything running during the instant of rewind.
The modern masters of the device make its operation part of the watch's character. François-Paul Journe's Tourbillon Souverain carries a one-second remontoire whose firing gives the seconds hand its precise, deadbeat tick — one of the most quietly addictive details in contemporary watchmaking. A. Lange & Söhne uses a one-minute remontoire in the Zeitwerk, where it doubles as the power source that snaps the jumping digital discs forward, and a refined version in the Lange 31, where a 31-day mainspring would otherwise deliver wildly different torque in week one and week four. Girard-Perregaux's Constant Escapement attacks the same problem at the final interface, storing energy in a buckling silicon blade that releases an identical snap of impulse every beat. All of these watches measure conspicuously well, and the constant-force machinery is a major reason.
The two prestige mechanisms are often confused because both are visible, expensive, and historically freighted. They address different errors: the tourbillon averages out positional error by rotating the escapement through gravity's directions; constant-force devices suppress torque-dependent error by evening the energy supply. A watchmaker chasing real-world rate stability gets more from the remontoire — which is why the most chronometrically serious modern movements pair a conventional escapement with constant force, and why several makers (Journe among them) combine both in one watch when making the full argument.
Why it matters for collecting
Constant-force machinery separates technically serious movements from adequate ones at the top of the market. A remontoire or fusee is a deliberate decision to spend size, cost, and complexity on rate consistency — a quality no one can see at dinner — and its presence in a movement is reliable evidence of engineering ambition rather than theatre. When specialists describe Journe or Lange movements as serious, this is part of what they mean. It is also a useful lens in reverse: a movement loaded with visual spectacle but indifferent to power delivery has told you what its priorities are.
The most elegant complication may be the one designed to make the rest of the watch behave less dramatically. A remontoire does nothing a buyer can show off at dinner. It simply makes every beat more consistent than the one before it — which is, after all, the entire original point of the machine.