Cartwright's Parallel Motion is a four-bar linkage that guides the end of a beam-engine piston rod along an approximately straight vertical line without a slide bar. Where Watt's parallel motion uses a pantograph of three links to copy the beam-end arc, Cartwright's design uses a simpler crossed-link arrangement to cancel the lateral component of the beam swing. It exists because early double-acting steam engines pull as well as push, so a chain or rope guide will not work — the rod must be held rigid in both directions. The result is a low-friction guidance method that kept beam engines running at 20-30 RPM with piston-rod side load under about 2% of working steam load.
Cartwright's Parallel Motion Interactive Calculator
Vary beam length, swing angle, and frame-pivot drop to see the uncancelled beam arc and the estimated piston-rod tracking error.
Equation Used
The calculator compares the sideways sweep a beam end would have if it simply followed its circular arc with the estimated rod-side tracking error caused by frame-pivot settlement. The worked geometry note states that a 2 mm frame-pivot drop produces about 1 mm of sideways rod shift on a typical 0.6 m beam half-length.
- Beam swing is the half-angle from the mid-stroke position.
- Pivot-drop error uses the article relation: 2 mm bracket drop gives about 1 mm rod shift on a 0.6 m beam half-length.
- Linkage proportions are similar to the example geometry.
- Calculated tracking error is a mid-stroke alignment estimate, not a full dynamic bearing-load model.
How the Cartwright's Parallel Motion Actually Works
The problem Edmund Cartwright was solving in the 1780s is straightforward. A working beam swings in an arc, but the piston rod below it has to move in a perfectly straight line up and down the cylinder. If you bolt the rod directly to the beam end the rod gets dragged sideways, the gland packing wears oval in 200-300 hours of running, and the cylinder bore scores. Watt patented his pantograph parallel motion in 1784. Cartwright — better known for the power loom — patented a competing linkage that achieves the same straight-line approximation with different geometry, using a pair of crossed links rather than Watt's nested arrangement.
The trick in any approximate straight-line linkage is that two curved motions cancel each other along one axis. In Cartwright's version, the beam-end traces a circular arc one way, and a secondary radius bar traces an arc the other way. Where the two link to the piston-rod attachment point, the lateral components cancel to within a few thousandths of an inch over the working stroke. That tracking error is what matters in service. If you build the linkage with link-length tolerances looser than about ±0.5 mm on a 600 mm beam, the cancellation drifts and you get sideways thrust on the gland — you'll see the rod-packing rope discoloured on one face inside a week of operation. If the pivot bushings wear past about 0.4 mm radial slop, the rod end starts to figure-eight slightly and the cylinder top bush develops a witness mark on one side.
The geometry is fixed by the beam length, the radius bar length, and the pivot position of the radius bar. Get those three right and the dead-centre alignment of the rod sits within ±0.3 mm of true vertical across the full stroke. Get any of them wrong and no amount of fitting will fix it — the rod will always pull sideways at one end of the stroke.
Key Components
- Working Beam: The main rocking lever pivoted at the entablature centre. On a typical Boulton & Watt-sized engine the beam is 6-7 m long, cast iron with a wrought-iron strap, and swings through about ±15° of arc. The half-end the rod attaches to traces a circular path the linkage must straighten.
- Radius Bar: A secondary link pivoted on a fixed point on the engine house frame. Its length is set so its swept arc, combined with the beam-end arc, cancels lateral motion at the rod attachment. Length tolerance is critical — typically ±0.3 mm on a 1.2 m bar.
- Connecting Link: The cross-link tying the radius bar to the beam end. Carries the geometric cancellation. Forged wrought iron in period engines, with hardened pin bushings rated for 2-4 MPa bearing pressure at the pivots.
- Piston Rod Attachment Pin: The point where the linkage delivers straight-line motion to the rod top. Must sit within ±0.3 mm of true vertical across the full stroke. Pin diameter typically 50-75 mm in mill engine sizes, ground to h7.
- Frame Pivot Bracket: The fixed anchor for the radius bar. Mounted to the engine house masonry or cast frame. Any settlement here throws the geometry — a 2 mm drop of the bracket pushes the rod sideways by roughly 1 mm at mid-stroke on a typical 0.6 m beam half-length.
Where the Cartwright's Parallel Motion Is Used
Cartwright's Parallel Motion appeared in a narrow window — beam engines from roughly 1787 through to the 1830s, when slide-bar crosshead guides became universal and made all parallel motion linkages obsolete for new construction. Where you find Cartwright's version today is in preserved engines, working museum installations, and restoration projects where the original linkage geometry must be reproduced rather than substituted. A skilled millwright can usually identify Cartwright versus Watt geometry on sight by counting the links and looking for the crossed configuration.
- Heritage steam preservation: The Crofton Pumping Station 1812 Boulton & Watt engine in Wiltshire uses a Watt parallel motion, but several smaller estate engines from the same period — including a recovered colliery winder at the Black Country Living Museum — use Cartwright-pattern linkages and need original geometry maintained during restoration.
- Textile mill restoration: Quarry Bank Mill at Styal retains rocking-beam pumping gear with period parallel motion linkages. Replacement bushings on the radius bar pivots must match original pin centres to within 0.5 mm or the rod alignment shifts.
- Maritime museum exhibits: Early paddle-steamer beam engines preserved at the SS Great Britain dockyard in Bristol use parallel motion linkages of the late-Cartwright pattern. Reproductions for working demonstration runs require the link-length ratios held within 0.2%.
- Educational kinematics: Kinematics teaching collections at universities like Cornell's Reuleaux Collection and Cambridge's Whipple Museum hold Cartwright-pattern demonstrators showing the crossed-link straight-line approximation alongside Watt and Roberts variants.
- Historic mine pumping: The preserved Cornish-style pumping engine at the Pleasant Hill Lane museum site retains a parallel motion of Cartwright derivation driving the main pump rod. Restoration requires the radius-bar pivot bracket re-anchored to its original masonry datum within ±1 mm.
- Working replicas: Scale working models built for events like the Internal Fire Museum of Power demonstrations use Cartwright linkages where the original engine type warrants — typically 1:8 scale, link tolerances scaled accordingly to ±0.05 mm.
The Formula Behind the Cartwright's Parallel Motion
What you actually need to compute when fitting or reproducing a Cartwright parallel motion is the lateral tracking error — how far sideways the piston-rod attachment drifts from a true vertical line as the beam rocks through its arc. This number tells you whether the gland will run cool or whether the rod will scrub the packing. At the low end of normal beam swing, around ±5° of arc, tracking error sits in the sub-millimetre range and the rod runs effectively true. At nominal full swing of ±15°, the error grows but stays inside the rope-packing's compliance budget. Push past ±20° — which happens if someone resets the valve gear wrong on a restoration — and the error climbs steeply because the linkage's cancellation is only a second-order approximation, not a true straight-line generator. The sweet spot for a Cartwright linkage is the swing range it was geometrically tuned for, typically ±12° to ±15°.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| εx | Lateral tracking error of the piston rod attachment from true vertical | m | in |
| Lb | Beam half-length from main pivot to rod attachment | m | ft |
| Rr | Radius bar effective length | m | ft |
| θ | Beam swing angle from horizontal | rad | rad |
Worked Example: Cartwright's Parallel Motion in a restored 1815 Cornish pumping engine
You are checking the parallel motion geometry on a restored 1815 Cornish-pattern beam pumping engine at a heritage water board site in the Peak District. Beam half-length from main pivot to piston rod attachment is 3.0 m, radius bar effective length is 1.2 m, and the engine's working stroke takes the beam through ±15° from horizontal. You need to know the lateral tracking error at the rod top so you can spec gland packing compliance and decide if the bushings need re-bored.
Given
- Lb = 3.0 m
- Rr = 1.2 m
- θnom = 0.262 (15°) rad
- θlow = 0.087 (5°) rad
- θhigh = 0.349 (20°) rad
Solution
Step 1 — at nominal full swing of ±15°, compute the tracking error:
Step 2 — at the low end of normal running, ±5° beam swing (idle pumping or barring over slowly):
That is essentially nothing — the rod runs dead true, the gland packing barely sees a side load, and you could run for years without measurable scoring. Most heritage engines spend a lot of time in this range during demonstration cycles.
Step 3 — at ±20°, which happens if the valve gear has been reset incorrectly or the eccentric is mistimed:
4.43 mm of sideways drift is not what the gland was designed for. You will see the packing rope blackened on one face within a single working day, and the cylinder top bush will pick up a witness mark on the loaded side. The cubic dependence on θ is what makes this dangerous — going from 15° to 20° is only a 33% increase in swing but more than doubles the tracking error.
Result
Nominal lateral tracking error at ±15° beam swing is 1. 87 mm — well inside the compliance budget of standard graphited rope packing and entirely acceptable for a working heritage engine. The range tells the story: 0.07 mm at idle ±5° is invisible, 1.87 mm at nominal sits in the sweet spot the linkage was geometrically tuned for, and 4.43 mm at ±20° is where the engine starts to damage itself. If you measure tracking error closer to 3 mm at nominal swing, the most likely causes are: (1) the radius bar pivot bracket has settled in its masonry by 1-2 mm, throwing the geometric cancellation off, (2) the radius bar itself has been replaced at some point with an incorrect length — check it against the original drawing or a surviving sister engine, or (3) the connecting link pin holes have worn oval and the effective link length has grown by a millimetre or two.
Choosing the Cartwright's Parallel Motion: Pros and Cons
Cartwright's Parallel Motion competed with Watt's pantograph linkage and, later, with the slide-bar crosshead that ultimately replaced both. Each approach guides a piston rod straight, but they differ sharply on accuracy, frame complexity, and how forgiving they are of build tolerances. For anyone deciding whether to reproduce a Cartwright linkage on a restoration or substitute a different mechanism, these are the dimensions that matter.
| Property | Cartwright's Parallel Motion | Watt's Parallel Motion | Slide-Bar Crosshead |
|---|---|---|---|
| Tracking accuracy at full stroke | ±2 mm typical on a 3 m beam | ±0.5 mm typical (pantograph cancels to higher order) | ±0.05 mm — limited only by guide bar wear |
| Operating speed range (RPM) | 10-30 RPM beam engine speeds | 10-40 RPM beam engine speeds | 30-300+ RPM, suits horizontal high-speed engines |
| Frame complexity | 3 links, simpler than Watt | 5 links including pantograph, more parts | 2 parallel guide bars, simplest mechanically |
| Tolerance sensitivity | ±0.5 mm on link lengths to keep tracking inside spec | ±0.3 mm — pantograph stacks tolerance errors | ±0.1 mm guide-bar parallelism |
| Maintenance interval (heritage operation) | Pin bushings re-bored every 5,000-8,000 hours | Pin bushings re-bored every 4,000-6,000 hours (more pivots) | Guide bars regrind every 10,000-15,000 hours |
| Typical era and application fit | 1787-1830s beam engines | 1784-1840s beam engines, paddle steamers | 1830s onward, all engine layouts |
| Build cost (modern reproduction) | Moderate — 3 forged links, 4 pivots | Higher — 5 links, 6 pivots, more setup work | Lower — 2 machined bars, 1 crosshead casting |
Frequently Asked Questions About Cartwright's Parallel Motion
Count the links and look at the geometry. Watt's linkage is a pantograph — you'll see a parallelogram of four links plus a connecting bar, with the rod attached partway along one side. Cartwright's is simpler: a beam-end link, a radius bar pivoted to the frame, and a connecting link between them, with no parallelogram. If the rod attaches at the apex where two links cross over each other rather than to a parallelogram corner, it's Cartwright. On a heavily restored engine where original castings are missing, check archive drawings — the patent dates differ by a few years and most preserved sites have records.
Thermal expansion of the cast-iron beam and wrought-iron links is the usual culprit. A 3 m cast-iron beam grows about 1.4 mm per 30°C temperature rise. If the beam and the radius bar are different materials — common on engines repaired over their service life — they expand at different rates and the geometric cancellation drifts. You'll see tracking error go from 1.5 mm cold to 2.5 mm at running temperature.
The fix on a heritage engine is matched materials. If you must mix, calculate the differential expansion at running temperature and adjust the radius bar pivot bracket position by that amount on a sliding mount.
If you are restoring an engine for static or working museum display, reproduce the original. The kinematic signature — how the linkage moves through its arc — is part of the historical record and substituting a different geometry destroys that. Visitors who know what they're looking at will spot it immediately.
If the engine is being rebuilt for sustained commercial use, which is rare but happens at a few sites, the Watt linkage tracks roughly four times more accurately for only one or two extra pivots. But honestly, if accuracy matters that much you should be installing a slide-bar crosshead and accepting that the engine is no longer historically correct.
Tracking error is the lateral drift averaged over the stroke. Uneven packing wear usually points to an asymmetric error — the rod drifts more one way than the other. The two common causes are: an out-of-square radius bar pivot bracket, where the bracket bolt circle isn't perpendicular to the beam plane, and a bent connecting link from a previous over-stroke event. Set a dial indicator against the rod at top and bottom dead centre and read the lateral position at each — a healthy Cartwright linkage shows symmetric drift around vertical, an asymmetric reading means the geometry is twisted in plan, not just in elevation.
For a Cartwright linkage the radius bar length is geometrically tied to the beam half-length and the desired straight-line range. The classic ratio used by Cartwright in his patent examples is roughly Rr = 0.4 × Lb, which gives near-optimum cancellation across ±15° of beam swing. On the 3 m beam half-length example that gives 1.2 m, which matches the surviving examples.
If you have the original pivot bracket location surviving on the engine house frame, that fixes the geometry — measure from the bracket pivot to the rod attachment point at mid-stroke and that distance is your radius bar length. Don't guess. A 5% error in radius bar length pushes tracking error above 3 mm and you'll be stripping the gland inside a month.
The formula assumes ideal pin-jointed links with zero clearance. Real bushings have radial play — typically 0.05-0.15 mm on a freshly bored pin and bush, growing to 0.3-0.5 mm before re-boring is needed. With four pivots in the linkage, that clearance stacks. At mid-stroke the load reverses direction in some pivots and the rod attachment point shifts laterally by the sum of clearances on the loaded pivots — typically 0.2-0.6 mm of motion that has nothing to do with the geometric tracking error.
This is why heritage engineers rate Cartwright linkages by combined tracking-plus-clearance error rather than the geometric calculation alone. If you need the geometric value, take measurements at slow barring speeds where load reversal is gentle — at running speeds inertial effects mask the geometry.
You can, but you shouldn't. The linkage works in any orientation kinematically, but Cartwright's pattern relies on link weight to maintain pivot loading in one direction — pivots that load-reverse with each stroke wear oval much faster, often three times faster than unidirectionally loaded ones. On a vertical beam engine the link weights help; on a horizontal layout the gravity vector cancels out and pivot wear accelerates.
Slide-bar crossheads were developed specifically because they handle horizontal layouts cleanly. If you're building a horizontal engine, use a crosshead. The only horizontal-Cartwright installations that have survived are demonstration models running a few hundred hours a year, where the wear rate is acceptable.
References & Further Reading
- Wikipedia contributors. Parallel motion. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
