A pendulum is a mass suspended from a pivot that swings under gravity, with a period determined almost entirely by its effective length. Unlike a balance wheel and hairspring — which uses elastic restoring force and is sensitive to temperature and amplitude — a pendulum uses gravity itself as the restoring force, giving exceptional long-term stability when the amplitude stays small. That stability is why every precision regulator from Riefler to Shortt used a pendulum, and why a well-built seconds pendulum still holds time to better than 0.1 second per day.
Pendulum (general) Interactive Calculator
Vary target period, local gravity, and swing amplitude to size the effective length and see the pendulum response.
Equation Used
The calculator rearranges the small-angle pendulum period equation to find the effective length L needed for a target complete oscillation period T at local gravity g. Beat time is T/2, and the amplitude tile estimates circular-error rate loss from finite swing angle.
- Small-angle pendulum model for the main length calculation.
- Effective length is measured from pivot to center of oscillation.
- Bob mass, air drag, pivot friction, and escapement impulse are ignored.
- Amplitude loss uses the first terms of the finite-amplitude period correction.
Inside the Pendulum (general)
A pendulum swings because gravity pulls the bob back toward the lowest point of its arc. The restoring torque is proportional to sin(θ), but for small swings sin(θ) ≈ θ, which is why the motion is approximately simple harmonic and why the period depends almost entirely on the effective length L and gravity g. Keep amplitude under about 2° peak-to-peak and circular error stays under 1 part in 10,000. Push it to 6° and you pick up a measurable rate change — this is the classic isochronism problem that limited cheap kitchen clocks for two centuries.
The escapement feeds tiny impulses to keep the bob swinging against air drag and pivot losses, but every impulse also disturbs the natural period. The art of a good clock is making those impulses as small and as symmetrical as possible — a deadbeat escapement on a 1-second pendulum delivers around 1 mJ per beat, just enough to overcome a Q factor of 10,000 to 12,000. If the suspension spring is too stiff, the effective length shifts and the clock runs fast. If the bob's centre of oscillation is not exactly at the marked rating point, regulation by the rating nut becomes nonlinear and you chase the rate up and down for hours.
Failure modes are mostly thermal and mechanical. Temperature swings stretch a plain steel rod by about 11 µm per metre per °C — on a seconds pendulum that's roughly 10 seconds per day per °C if uncompensated, which is why gridiron, mercury, and Invar pendulums exist. Worn suspension springs develop a kink at the flexure line and the pendulum starts swinging out of plane, scrubbing the crutch and hunting the rate. A bob with a loose rating nut drifts seasonally as humidity changes the wood case dimensions around the suspension cock.
Key Components
- Bob: The mass concentrated at the lower end. For a seconds pendulum the bob is typically 2–7 kg of brass or lead, lens-shaped to reduce air drag. The mass itself does not change the period — only the centre-of-oscillation distance does — but a heavier bob raises Q by improving the energy-to-loss ratio.
- Rod or suspension: Carries the bob and sets the effective length. Steel rods drift roughly 11 µm/m/°C; Invar drops this to about 1.2 µm/m/°C. The rod must stay straight to within 0.1 mm over 1 metre — any bow shifts the centre of oscillation and changes the rate.
- Suspension spring: A thin spring strip — typically 0.1 mm thick blued steel — that flexes at the top and defines the pivot axis. A worn or kinked spring shortens effective length unpredictably and is the single most common cause of erratic rate in long-running clocks.
- Crutch and impulse pin: The link from the escapement to the pendulum. Lateral play above 0.05 mm here introduces side-shake that adds noise to the impulse and degrades isochronism.
- Rating nut: Threaded nut under the bob that raises or lowers it to fine-tune the period. On a seconds pendulum, one full turn of an M5 × 0.5 nut shifts rate by roughly 30 seconds per day — set the bob at the marked centre of oscillation before relying on the rating nut.
- Temperature compensation: Gridiron (steel and brass rods in opposition), mercury jar bobs, or Invar rods. A well-built mercury pendulum holds rate to 0.05 s/day across a 10°C swing; an uncompensated steel rod over the same swing drifts about 100 s/day.
Real-World Applications of the Pendulum (general)
Pendulums show up wherever you need a stable, low-cost time reference and you can give the device a fixed vertical orientation. They cannot work on a moving platform — that ruled them out for marine chronometers — but for tower clocks, regulators, scientific instruments, and gravity measurement they remained the gold standard from 1656 until quartz arrived in the 1930s. Even today, a properly built precision pendulum will outperform many quartz movements on long-term drift if you control the temperature.
- Public timekeeping: Tower clocks like the Great Clock of Westminster (Big Ben), with a 4-metre, 300 kg pendulum beating 2 seconds, regulated by adding or removing old penny coins to shift the centre of oscillation.
- Scientific regulators: The Shortt-Synchronome free pendulum clock used at the Royal Greenwich Observatory from 1924, holding time to about 1 second per year using a master pendulum in a vacuum tank.
- Domestic horology: Vienna regulators and longcase clocks with seconds pendulums, where the rod length sits at 994 mm to give a 2-second period (1-second beat) at 45° latitude.
- Geophysical instruments: Kater's reversible pendulum, used from the 1820s into the 20th century to measure local gravity g to 5 decimal places.
- Seismology: Long-period horizontal pendulums in early seismographs such as the Milne-Shaw, where periods of 10–20 seconds detect distant earthquake waves.
- Engineering education: Foucault pendulum demonstrations like the 67-metre installation at the Panthéon in Paris, showing Earth's rotation.
The Formula Behind the Pendulum (general)
The small-angle period formula tells you how long a single swing takes given the pendulum's effective length and local gravity. At the low end of practical clock pendulums (around 250 mm, beating roughly half-second), the rate is highly sensitive to length errors — a 0.1 mm length change shifts a clock by about 17 s/day. At the nominal seconds pendulum length of 994 mm the same 0.1 mm error costs you only 4 s/day, which is the sweet spot every regulator builder targets. Push beyond 4 metres into tower-clock territory and length sensitivity drops further, but air drag, flex of the rod, and amplitude variation start to dominate the error budget instead.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| T | Period of one complete oscillation (out and back) | s | s |
| L | Effective length from pivot to centre of oscillation | m | ft |
| g | Local gravitational acceleration | m/s² | ft/s² |
| π | Mathematical constant ≈ 3.14159 | dimensionless | dimensionless |
Worked Example: Pendulum (general) in a precision seconds pendulum for a regulator
A small instrument workshop in wellington new zealand is building a precision seconds pendulum for a deadbeat-escapement regulator at latitude 41° south, where local gravity is g = 9.8030 m/s². The target period is T = 2.000 s (a 1-second beat). The bob is a 5 kg lens of cast lead, the rod is Invar to keep thermal drift under 1 s/day across a 10°C lab swing. Workshop bench temperature is controlled to 20 ± 1 °C. You want to know what rod length to cut, and how the rate behaves if the apprentice cuts it 1 mm long or 1 mm short.
Given
- T = 2.000 s
- g = 9.8030 m/s²
- Bob mass = 5 kg
- Rod material = Invar (α ≈ 1.2 × 10⁻⁶ /°C) —
Solution
Step 1 — rearrange the period formula to solve for nominal length L:
Step 2 — plug in the Wellington gravity value and target period:
So the centre of oscillation must sit 993.1 mm below the suspension flex point. That is the number you cut to, then trim with the rating nut.
Step 3 — check the low end of the realistic length tolerance, 1 mm short at L = 992.1 mm:
That 1 millisecond per swing sounds tiny, but on a clock running 43,200 swings per day it accumulates to roughly 43.6 seconds per day fast — a clock that gains over half a minute by tea time. Most apprentice builds end up here.
Step 4 — high end, 1 mm long at L = 994.1 mm:
Roughly 43 seconds per day slow. The asymmetry is small because at this length-to-error ratio the formula is nearly linear, but the lesson is sharp — a millimetre matters. The rating nut on a typical seconds pendulum gives you about ±60 s/day of trim authority before you run out of thread, so a 1 mm cutting error is recoverable but eats half your adjustment range.
Result
The nominal cut length is 993. 1 mm from suspension flex point to bob centre of oscillation, giving the desired 2.000 s period at Wellington gravity. In practice that means a clock that beats once per second to within a few seconds per day before any further regulation. At 1 mm short the clock gains around 43 s/day; at 1 mm long it loses around 43 s/day — the sweet spot is to cut 0.5 mm long deliberately and bring it down with the rating nut, since adding length is harder than removing it. If your measured rate is more than 60 s/day off predicted, the most likely culprits are: (1) the marked centre of oscillation on the bob is not where you assumed it is — lens-shaped lead bobs often have their COO 2–4 mm above the geometric centre, (2) the suspension spring has a permanent set or kink shortening effective length by 0.1–0.3 mm, or (3) the apprentice measured to the top of the suspension block instead of the actual flex line, which on a typical Vienna-style cock sits 1.5 mm below the clamp face.
When to Use a Pendulum (general) and When Not To
A pendulum is one of three classical time references — pendulum, balance wheel, and quartz crystal — and the choice between them comes down to environment, accuracy target, and budget. Pendulums dominate stationary precision and large-scale public timekeeping; balance wheels dominate portable mechanical timekeeping; quartz dominates everything where you need cheap, accurate, and electrical.
| Property | Pendulum | Balance wheel & hairspring | Quartz oscillator |
|---|---|---|---|
| Typical accuracy (well-regulated) | 0.01–1 s/day | 2–10 s/day | 0.1–1 s/day |
| Q factor | 8,000–25,000 (vacuum: 100,000+) | 100–300 | 10,000–100,000 |
| Works on moving platform | No - must be vertical and stable | Yes — designed for it | Yes |
| Sensitivity to temperature (uncompensated) | ~10 s/day/°C (steel rod) | ~10 s/day/°C | ~0.05 s/day/°C |
| Power input per beat | ~1 mJ (seconds pendulum) | ~1 µJ | ~1 nJ |
| Typical service life before rebuild | 50–150 years | 5–10 years | 10–30 years (battery limited) |
| Build cost (precision grade) | Medium — Invar rod, mercury bob | High — jewelled bearings, hairspring matching | Low — mass-produced crystal |
Frequently Asked Questions About Pendulum (general)
Invar handles the rod, but the suspension spring is usually plain blued steel and its modulus drops about 240 ppm per °C as it warms. A softer spring effectively lengthens the pendulum slightly and should slow the clock — but in practice most builders see the opposite, a gain, because warmer air is less dense, drag drops, amplitude rises, and circular error makes the clock run fast.
Check amplitude with a protractor card behind the bob. If it has climbed from 1.5° to 2.5° peak between the two rooms, that's your culprit, not the rod.
Cost and target accuracy decide it. Invar is the easiest — drop in a single rod, accept about 1.2 ppm/°C residual drift, and you'll hold under 1 s/day across normal room swings. Mercury jar bobs are slightly better thermally (you can null the drift by adjusting mercury column height) but mercury is now heavily restricted in most jurisdictions and the bob is fragile.
Gridiron is the right choice when you are restoring a period clock and authenticity matters, or when you want a fully mechanical compensation that needs no exotic alloys. Expect to spend a day adjusting brass-to-steel rod length ratios to actually null the drift — the textbook 5:9 ratio is a starting point, not a final answer.
Two effects, both real. First, opening the door drops air pressure briefly as room air mixes in, changing buoyancy on the bob — for a 5 kg lead bob the effect is small, around 0.1 s/day per 10 mbar. Second and bigger, you change the air currents around the bob, which alters drag and amplitude.
If you see a 2–5 second per day step every time the case opens, that's the amplitude effect. The fix is a tighter case seal or, on a serious regulator, running the pendulum in a partial vacuum like the Shortt clocks did.
Almost always the suspension spring. A new spring flexes on a single line; a worn or kinked spring flexes on a slightly twisted line, and the pendulum precesses around it. You'll see the bob tracing a narrow ellipse instead of a clean line.
Pull the spring and inspect under a 10× loupe. Any visible kink, pitting, or rust spot means replace it. While you're in there, check that the crutch fork has under 0.05 mm of side-shake on the impulse pin — excess play here also drives out-of-plane swing.
The fractional period error from circular error is approximately θ²/16 in radians. At 2° peak amplitude (0.0349 rad) that's about 76 ppm, or 6.6 s/day — significant for a regulator. At 1° peak it drops to 1.6 s/day. At 0.5° peak you're below 0.5 s/day and circular error is no longer your dominant error term.
That's why precision regulators run at very small amplitudes — the Riefler clocks held about 1.5° peak, the Shortt master pendulum ran near 1°. If your hobby clock runs at 4° or more, you're fighting circular error before anything else.
You have backlash or thread slop. The bob is sitting on the nut by gravity, but if the nut threads are worn or the bob's central hole is sloppy on the rod, the first half-turn just takes up clearance without lifting the bob. Once contact is made, the rate moves as expected.
Check by marking the nut with a pencil line, turning a quarter turn, and seeing if the bob actually moves on the rod with calipers. If it doesn't, you need a new rating nut or a tighter rod-to-bob fit. A correctly fitted M5 × 0.5 rating nut on a seconds pendulum should give a smooth ~15 s/day per quarter turn with no dead zone.
References & Further Reading
- Wikipedia contributors. Pendulum. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
