A Fusee Chain and Spring Drum is a horological torque-equalising mechanism that links a cylindrical spring barrel to a tapered cone (the fusee) by a fine articulated chain. John Harrison's H1 marine chronometer of 1735 used one to keep rate stable as the mainspring unwound. As the spring weakens, the chain pulls from a larger fusee radius, raising leverage to compensate. The result is near-constant torque at the centre wheel across an 8-day run — typically holding rate within ±2 seconds per day where a plain barrel drifts 30+.
Fusee Chain and Spring Drum Interactive Calculator
Vary spring torque, fusee radius, drum radius, and efficiency to see chain tension and equalized output torque.
Equation Used
The spring drum creates chain tension from the mainspring torque. The fusee converts that chain tension back into output torque using its local radius. As the spring weakens, a larger fusee radius keeps the output nearly constant.
- Spring drum radius is constant.
- Chain tension equals spring torque divided by drum radius.
- Fusee output torque is chain tension times fusee radius.
- Efficiency accounts for chain, bearing, and groove losses.
Operating Principle of the Fusee Chain and Spring Drum
The Fusee Chain and Spring Drum, also called the Fusee chain and spring-box (watch) in pocket-watch literature, solves one stubborn problem — a coiled mainspring delivers far more torque when fully wound than when nearly run down. In a plain going barrel that torque variation hits the escapement directly, and rate sags as the spring unwinds. Drop a tapered fusee between the spring barrel and the centre wheel, link the two with a 200+ link articulated chain, and you flip the leverage relationship so the geometry cancels the spring's weakness.
Winding pulls the chain off the spring drum and wraps it onto the fusee, starting at the small end of the cone. As you wind, the chain climbs toward the large end. Running, the action reverses — the chain feeds back onto the drum, and the fusee is driven from progressively larger radii as the spring weakens. The fusee curve is not a simple cone — it's a logarithmic profile cut so that the product of mainspring torque × fusee radius stays constant. Get that curve wrong by even 0.2 mm in radius at any turn and rate will hump or dip across the run; we have measured 1820s English bracket clocks where a re-cut fusee shifted daily rate from +18 s to +2 s.
What goes wrong, in order of frequency: chain stretch (the riveted links elongate after decades, throwing the leverage match off), maintaining-power click failures during winding (the clock stops while you wind because the going train loses drive), and chain breakage at the hooked ends — almost always at the spring-barrel hook because that's where peak tension lives at full wind. A fusee chain at the wrong pitch will not seat in the fusee groove, jumps, and snaps within hours.
Key Components
- Spring drum (mainspring barrel): A closed cylindrical box housing the coiled mainspring. Typical pocket-watch drums run 18-22 mm diameter with a steel spring 0.10-0.18 mm thick and 300-400 mm long. The drum is fixed on its arbor — it does not rotate freely as a going barrel does, because the chain anchors to its outer wall.
- Fusee cone: A truncated logarithmic cone, typically 12-20 mm at the large diameter and 6-9 mm at the small. Cut with a helical groove of pitch matching the chain (commonly 0.30-0.45 mm pitch for watches, 0.7-1.2 mm for clocks). The curve must equalise torque to within 5% across the run.
- Fusee chain: An articulated chain of riveted brass or steel links — pocket-watch chains use 0.25-0.35 mm wide links and run 150-250 links long. Tensile rating sits around 8-15 N. Made historically by women and children in Christchurch, Dorset until the 1920s — each link hand-formed.
- Maintaining power spring: A small auxiliary spring (Harrison's invention, 1750s) that drives the going train for the few seconds the fusee is being wound backward. Without it, the clock stops during winding. Typically delivers 30-60 seconds of reserve drive.
- Stopwork (fusee stop iron): A pivoted finger that drops into a slot at the top of the fusee to prevent overwinding. Critical — the chain breaks instantly if you wind past full, and almost every dead 18th-century fusee watch failed this way at some point in its life.
- Hooks and posts: The chain terminates in small steel hooks engaging posts on the drum and the fusee. The fusee hook sees peak tension at full wind, the drum hook sees peak tension at run-down. Hook fatigue is the most common chain failure point.
Where the Fusee Chain and Spring Drum Is Used
The fusee solved the rate-stability problem for two centuries before the going barrel with a long, weak mainspring eventually replaced it for most production watches. You still find fusees today wherever rate matters more than thinness or cost — marine chronometers, high-grade English bracket clocks, and a small revival in modern haute horlogerie.
- Marine navigation: John Harrison's H1 (1735) and H4 (1759) marine chronometers — the H4 used a Fusee Chain and Spring Drum and held rate within 5 seconds across the 81-day Atlantic voyage, winning the Longitude Prize.
- English clockmaking: Thwaites & Reed bracket clocks, 1780s onward — virtually every quality English fusee bracket clock built between 1700 and 1900 used this mechanism, including pieces by Thomas Tompion and George Graham.
- Pocket watch manufacture: English fusee verge and lever pocket watches (1700-1900) — the Fusee chain and spring-box (watch) configuration was standard in English work long after Swiss makers had moved to going barrels.
- Modern haute horlogerie: A. Lange & Söhne Tourbograph Pour le Mérite (2005) and Breguet Tradition 7047 — both use a miniaturised fusee and chain to drive a tourbillon at constant torque.
- Astronomical regulators: High-grade observatory regulators by Dent and Frodsham used fusees on shorter 1-day runs where rate stability beat the convenience of an 8-day going barrel.
- Turret and tower clocks: Some 18th- and 19th-century turret movements used heavy fusees with iron chains to even out the long mainspring runs before weight-drive became standard.
The Formula Behind the Fusee Chain and Spring Drum
The governing relationship is the torque-equalisation equation — the product of mainspring torque and fusee radius at any instant must equal the (constant) torque delivered to the going train. At the low end of the typical operating range (chain near the small end of the fusee, spring nearly run down), the fusee leverage is at its minimum and the spring is at its weakest — both factors must multiply to the same target. At the high end (full wind, chain at the large end of the fusee), the fusee radius is small but spring torque is at maximum. The sweet spot sits in the middle of the run where curve cutting tolerances are most forgiving — errors in fusee profile show up worst at the two ends, which is why old chains that have stretched produce rate humps near full wind and rate sag near run-down.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Tout | Torque delivered to the centre wheel (target constant) | N·mm | oz·in |
| Tspring(θ) | Instantaneous mainspring torque as a function of turns wound | N·mm | oz·in |
| rfusee(θ) | Effective fusee radius at the chain departure point | mm | in |
| rdrum | Constant spring drum radius (chain wraps a cylinder) | mm | in |
| θ | Turns wound from fully run-down state | rev | rev |
Worked Example: Fusee Chain and Spring Drum in an 1820s Thwaites bracket clock fusee verification
A heritage furniture restoration studio in Adelaide brings in an 1820s English fusee bracket clock attributed to Thwaites for a torque audit before re-mounting in a colonial mahogany case. The mainspring delivers 80 N·mm at full wind and 32 N·mm at run-down across 7 turns. The spring drum radius is 9.0 mm. You need to verify that the existing fusee profile equalises output torque, and check the practical range of behaviour at the start, middle, and end of the 8-day run.
Given
- Tspring,full = 80 N·mm
- Tspring,down = 32 N·mm
- rdrum = 9.0 mm
- θtotal = 7 rev
- Tout,target = 20 N·mm
Solution
Step 1 — at the high end of the run (full wind, θ = 0), solve for the required fusee radius. The chain pulls from the small end of the cone here, so rfusee must be small to keep Tout from spiking:
That 2.25 mm is the effective radius at the chain departure point near the top of the fusee. In practice, the cone diameter at this point would be roughly 5-6 mm including the groove depth.
Step 2 — at the nominal mid-run point (θ = 3.5 turns, Tspring ≈ 56 N·mm by linear interpolation):
This is the geometry the clock spends most of its life at — and it's where worn or stretched chains cause the smallest rate error. A 0.05 mm chain stretch here shifts Tout by about 1.5%, barely measurable on a daily rate.
Step 3°— at the low end (run-down, θ = 7 turns, Tspring = 32 N·mm). The fusee must now pull from a much larger radius to compensate:
So the cone profile must rise from 2.25 mm to 5.625 mm effective radius across 7 turns — a 2.5× ratio. If you measure a real fusee and the large-end radius is only 4.8 mm instead of 5.625 mm, the clock will lose roughly 12-15 seconds per day in the last 24 hours of the run as torque to the escapement collapses.
Result
The fusee profile must sweep from 2. 25 mm effective radius at full wind to 5.625 mm at run-down to hold the centre wheel at a constant 20 N·mm. In practice this means rate stability inside ±2 seconds per day across the 8-day run instead of the 25-40 second drift a plain going barrel would produce on the same spring. The mid-run sweet spot is forgiving — small chain wear shows up first as a rate dip in the final 36 hours, before any error is visible at full wind. If your measured rate sags more than 8 seconds per day late in the run, the three usual culprits are: (1) chain stretch lengthening effective lever at the small end of the fusee, (2) a worn or incorrectly re-cut large-end groove giving sub-spec leverage, or (3) mainspring set — the spring has lost amplitude and now delivers under 28 N·mm at run-down instead of the design 32 N·mm. Replace or re-cut in that order.
When to Use a Fusee Chain and Spring Drum and When Not To
The fusee competes with three other torque-management strategies. The going barrel is simpler and cheaper. The constant force escapement (remontoire) does the same job at the escapement end of the train. The stackfreed — a 16th-century cam-and-spring scheme — predates the fusee and is now obsolete. Picking between them comes down to how much rate stability you actually need and what room you have to fit the cone.
| Property | Fusee Chain and Spring Drum | Going Barrel (plain) | Remontoire (constant force) |
|---|---|---|---|
| Rate stability across full run | ±2 s/day typical | ±25-40 s/day | ±0.5 s/day |
| Mechanical complexity (part count) | High — 200+ chain links plus stopwork | Low — single barrel and click | Very high — auxiliary winding train |
| Vertical space required in movement | High — fusee cone adds 12-20 mm height | Low — flat barrel only | Medium — auxiliary spring assembly |
| Build cost (modern reproduction) | £800-2,500 for chain alone | £40-150 for barrel and spring | £2,000-8,000 fully fitted |
| Service interval before chain or spring degradation | 20-40 years typical | 8-15 years (mainspring fatigue) | 5-10 years (auxiliary winding wear) |
| Best fit application | Marine chronometers, high-grade clocks | Mass-produced wristwatches and clocks | Observatory regulators, tourbillons |
| Failure mode if neglected | Chain breakage at hook | Mainspring set, gradual rate loss | Auxiliary train jam stops the clock |
Frequently Asked Questions About Fusee Chain and Spring Drum
Almost always pitch mismatch. The fusee groove was cut for the original chain pitch — typically 0.35 mm on English work — and a modern replacement chain at 0.32 or 0.40 mm pitch will not seat correctly. The chain rides high in the groove, the effective fusee radius shifts unpredictably across the run, and you get rate humps at random turns. Verify pitch with a 10× loupe and a pin gauge before fitting. If the chain doesn't seat dead in the groove with no rocking, send it back.
Second cause: chain length. A chain even 3-4 links too long will not pull tight at full wind, leaving slack that snaps when the stopwork engages.
Plot daily rate across the full run. Chain wear produces a smooth, monotonic rate decline toward the end of the run because stretch accumulates linearly with use. A miscut fusee curve produces a non-monotonic signature — typically rate humps or dips at specific turns matching where the cutter ran off-profile.
Rule of thumb: if rate is fine for 5 days then collapses, suspect chain or spring. If rate wobbles ±10 s/day from day one but in a repeatable pattern that returns to the same shape after each rewind, the fusee profile itself is wrong.
For a longcase (weight-driven) you don't need either — gravity gives constant torque for free. For a spring-driven build, the choice depends on your rate target. If you'll accept ±15 s/day, a long going barrel with a sub-rated spring (using only the middle 60% of its travel) is far cheaper and simpler. If you want sub-±5 s/day or you're building to a heritage spec, the fusee earns its complexity.
Modern fusee chains from specialist makers run £800-1,500 alone, before the cone work — budget accordingly.
The maintaining-power spring is either weak, dirty, or the click is binding. When you wind, the going train must be driven by the auxiliary spring for those 10-30 seconds. If that spring has lost set (common after 50+ years), or the click ratchet sticks because of dried oil, the train loses drive momentarily.
Diagnostic: wind very slowly while listening to the escapement. If the tick stops within 2-3 seconds of winding starting, the maintaining power is failing. Re-tension or replace the spring and clean the click pivot — don't just oil it, the old grease has to come out.
No — and the failure mode is dangerous. Fusee chains are articulated in two planes so they can wrap a cone with continuously varying radius. Bicycle and roller chains articulate in one plane only and will bind, jump the groove, and snap under spring tension. A snapping fusee chain at full wind delivers the entire mainspring's stored energy through the broken end, often damaging the centre wheel pivots or the dial side of the movement.
The only correct replacement is a properly made fusee chain at the original pitch from a specialist supplier such as P. P. Thornton or one of the few remaining English chain makers.
On a typical English bracket clock chain of 200 links at 0.35 mm pitch, expect 0.3-0.8 mm total elongation across 50 years of regular use. That sounds trivial but it shifts the chain's departure point on the fusee by roughly half a turn at run-down, which is where the curve is steepest. We've measured original 1820s chains that pushed late-run rate from the design ±2 s/day to -18 s/day on day 7, purely from chain stretch.
That's why a chain replacement usually transforms a tired fusee clock — even before any other work.
References & Further Reading
- Wikipedia contributors. Fusee (horology). Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
