Driving torque is the twisting force that the mainspring transmits to the gear train and, ultimately, to the regulating organ. It is not a fixed quantity: it changes continuously as the spring unwinds, and this variation is the central problem in mechanical watchmaking. A new caliber, freshly wound, performs well not because it is more precise in absolute terms, but because the torque is at its maximum and the balance wheel oscillates with full amplitude. As the spring unwinds, the torque diminishes, the amplitude drops, and precision degrades. Everything that separates research watchmaking from ordinary watchmaking is, in substantial measure, how this problem is addressed.
There are two great historical families of solutions. The first acts on the energy source: multiple barrels in parallel, longer and thinner springs, self-regulating barrels. Parallel barrels — two, four as in the Singer Calibre 4 — do not necessarily multiply the power reserve, but flatten the torque curve: the force delivered remains more constant throughout the entire discharge because the peaks and dips of individual springs compensate for each other. A thinner spring accumulates energy over more turns with less tension per turn, reducing friction and making the release more linear. The second family acts downstream of the source: fusée-and-chain, maltese cross stop-work, remontoir d'égalité, Gravity Equal Force systems. The fusée is the oldest and most elegant mechanism: a chain links the barrel to a conical pulley, and the pulley's geometry compensates for the drop in torque by increasing the lever arm as the spring unwinds. The maltese cross stop-work cuts off the beginning and end of the discharge — the zones of maximum irregularity — and forces the movement to work only in the central band of the spring, where the torque is most stable. The remontoir d'égalité goes further: it recharges at regular intervals a small auxiliary spring that transmits the impulse to the balance wheel, completely separating the regulating organ from the fluctuations of the barrel.
From the bench I see a recurring error: confusing power reserve with torque stability. A watch with 80 hours of power reserve and variable torque is chronometrically less reliable than one with 42 hours and a flat curve. The torque indicator — which draws information directly from the barrel and makes it visible on the dial — makes sense precisely because residual charge is not synonymous with constant performance. Seiko's Spring Drive adds an interesting variation with the Torque Return System, which recovers energy from the electromagnetic brake of the glide spring and returns it to the system: I do not know enough details of the 9R02 caliber to describe its precise mechanics, but the recovery principle is documented. What I know with certainty, after fifty years at the bench, is that a caliber with a flat torque curve adjusts better, ages better, and maintains chronometric performance over time in a way that a conventional caliber cannot match.