The everlasting staircase: how the treadmill learned to measure you

In October 1858, a London physician walked into Coldbath Fields prison, climbed onto the treadwheel alongside the convicts, and started breathing into a tube.

His name was Edward Smith. He was not serving a sentence. He had come to measure.

Smith brought a chemist with him — Edward Frankland, one of the more formidable analysts in Britain — and between them they collected the air coming out of the men on the wheel, and out of Smith himself, and worked out how much had been converted into carbonic acid. He had run a version of the trial the year before, and reported the results to the Royal Society and to the newly founded National Association for the Promotion of Social Science. As far as the record shows, he was the first person to take proper quantitative measurements of a human being while that human being was exercising.

He did it on a punishment machine, and that was not a coincidence. In 1858 the prison treadwheel was very nearly the only apparatus in Britain capable of imposing a known, repeatable, adjustable quantity of work on a person for hours at a stretch. If you wanted to study effort, you went where effort was being administered by the state.

The machine in your gym is descended from that wheel. Not by rebranding. By inheritance of purpose.

The engineer was solving a paperwork problem

Around 1818 — the dating is soft, and I will come back to that — a Suffolk magistrate asked an Ipswich engineer named William Cubitt to do something about the prisoners at Bury St Edmunds gaol. The complaint was not that they suffered too little. It was that they were idle, that the law had begun to demand "hard labour" of them, and that nobody could say what hard labour was.

Cubitt's answer was a machine. He took the treadwheel — an old device for lifting things, worked from the inside like a hamster wheel — turned it inside out, stretched it long, and put the steps on the outside. Prisoners stood in a row along a rotating drum and climbed. Partitions kept them from leaning on a neighbour; a cover over the top kept them stepping near the axle, where their weight did the most good.

It looked like a staircase that never ended, and that is what people called it: the everlasting staircase.

But look at what Cubitt had really built. Not an instrument of torture. A work meter. The steps were a fixed height. The revolutions were counted. The number of men on the drum was known. For the first time, a prison governor could write down, in figures, precisely how much labour a sentence had extracted from a body.

Before the wheel, effort was a moral quantity

This is hard to feel from here, because we have had numbers so long we forget they were installed.

Before Cubitt, hard work was a judgement. You could say a man had worked hard, and you could disagree with someone who said he hadn't, and there was no third thing to appeal to. The wheel supplied the third thing. It converted a moral category into an arithmetic one.

Simon Schaffer, the Cambridge historian of science, argues in a 2025 paper in History of Science that this is the treadmill's genuine significance. The prison wheels became experimental sites, and the nineteenth-century sciences of energy and of labour power were built partly on top of them. The machinery of punishment did real work in the formation of physiology and physics.

Read that the other way round, because it inverts the story most of us have heard. The gym machine did not descend from the punishment machine by shedding its cruelty. It descended from it by keeping the measurement.

Grinding the wind

Here is where it turns strange. Many wheels were connected to something real: a corn mill, a water pump. Forty-four prisons in England used theirs to grind grain. Others were connected to nothing whatsoever — the drum turned a set of vanes that beat against the open air, and the sole function of those vanes was to make the climbing harder.

The men at Coldbath Fields had a name for this. They called it grinding the wind.

You can hear the contempt in it. You can also hear, if you listen a moment longer, the entire operating logic of modern exercise equipment. Resistance for its own sake. Effort that produces nothing at all except the fact of having been expended. Every time you thumb the incline up on a treadmill you are grinding the wind, and the man who coined the phrase would have recognised exactly what you were doing and would not have been impressed by it.

The difference is consent. That is nearly the whole moral distance between a prison and a gym. But mechanically, it is the only one.

Three things the popular version of this story gets wrong

1. It was not designed as torture. Cubitt's brief was idleness and the administrative problem of proving a sentence had been served. The cruelty was real and it arrived quickly, but it was a consequence of the design, not the specification.

2. It was not "rebranded" as fitness a century later. Physiologists were using prison wheels as laboratory apparatus while those wheels were still punishing people. Smith's trials ran in 1857 and 1858; abolition came decades afterwards. The two careers overlapped.

3. The tidy dates are not tidy. Sources put Cubitt's design at 1817 or 1818 and the first installations at 1819, and estimates of the daily climb range from roughly 6,600 to 17,000 vertical feet depending on the prison. Anyone quoting one crisp number has quietly chosen it.

TWO CENTURIES, ONE FUNCTION

The unit kept changing. The job never did.

The machine, and who was on itWhat it made of you

1. 1818Cubitt’s everlasting staircaseBury St Edmunds and Brixton gaolsRevolutionscounted by the governor
2. 1857Edward Smith climbs on with a tubeColdbath Fields, with the chemist FranklandCubic inches of airthe first exercise measurements
3. 1889Zuntz motorises the belt, Berlinthe motor sets the pace, not the manGas exchangeoxygen in, carbon dioxide out
4. 1902The last prison wheels come downabolished by the Prison Act 1898—the measuring outlives the punishment
5. 1963Bruce stages the treadmill testthree-minute stages, ECG throughoutMinutestime on the belt becomes the score
6. todayAn algorithm on the wristheart rate against pace, modelledmL/kg/minestimated — no oxygen is measured

The doctor who volunteered for the wheel

Back to Edward Smith, the hinge of the whole story.

In February 1857, at the Duke of York's column beside St James's Park, Smith measured respiration and estimated an excess of some 600 cubic inches of air moved on the discipline mill. In October 1858 he repeated the trials at Coldbath Fields with Frankland, sampling the gases going in and coming out.

What he was actually chasing was food. Prisoners on the wheel were fed a ration set by committee, and Smith wanted to know whether the ration matched the work. So he sorted human beings by how much air they moved — unoccupied gentlemen, ordinary tradesmen, hardworking labourers — and showed that the men on the wheel belonged firmly in the last category and were not being fed as though they did.

The Dictionary of Scientific Biography is unambiguous about what this amounted to: Smith was the first to devise quantitative methods suitable for studying the human being during exercise, and his data on breath volume, pulse and carbon dioxide production at graded levels of work underwrote much of the muscular-exercise research that followed.

There is a further twist. Smith's numbers indicated that starch and fat, rather than protein alone, were the principal fuel for muscular work — a direct challenge to the theory of the day, and a result that fed into the emerging account of the conservation of energy. The modern, unremarkable fact that you burn carbohydrate and fat in order to move was established, in part, by a doctor standing on a prison wheel with a tube in his mouth.

What it cost, which the fitness version never mentions

None of this redeems the wheel, and the story is better untidied.

Prisoners typically spent around six hours a day on the drum, in quarter-hour shifts with five-minute breaks. The vertical climb depended on the prison, and the surviving estimates disagree by a factor of more than two. Whatever the true figure, it was a great deal of work performed by underfed men on a machine with no brake.

The harms are in the record. Hernias. Rheumatism. Damage to the lungs. Nursing women who lost their milk supply. Badly built wheels that took a limb, or a life, when a foot missed a step. From the 1820s the medical journals argued about it in public — Benjamin Hutchinson published "Observations on the Tread-Mill" in The London Medical and Physical Journal in August 1823 — and by the 1830s opinion had swung against the machine.

The economics collapsed at roughly the same time. Industrial power made human beings a ridiculous way to grind grain, so the wheels that had at least been useful stopped being so. What remained was the punishment and the arithmetic.

The Prison Act of 1865 entrenched it anyway, requiring male prisoners over sixteen sentenced to hard labour to spend time on a wheel or a crank. The Prison Act of 1898 abolished it. There were 39 treadmills in English prisons in 1895; by 1901, 13; by 1902 they were gone.

How many people were injured across eighty years, nobody knows. No reliable count was kept — which is its own fact about what the institution was measuring and what it wasn't.

The belt replaces the wheel

Berlin, 1889. Nathan Zuntz, a veterinarian and physiologist at the Royal Agricultural College, built with his collaborator Lehmann the first motor-driven treadmill — a Laufband, a running belt. Zuntz already had the respiration apparatus he had developed with Geppert in 1886 for analysing exhaled air. What he lacked was a way to set the pace himself rather than let the subject set it.

That inversion is the quiet turning point of the whole history.

On Cubitt's wheel the drum was turned by the men on it; they had to keep up, but it moved because they moved. On Zuntz's belt a motor decided the speed and the human complied. The measurement got cleaner. The subject lost the last of their say. Every treadmill since has worked Zuntz's way, which is why one feels different from a road even when the numbers match.

The apparatus was shown at the Dresden hygiene exhibition in 1911. Zuntz added an X-ray rig in 1914 to watch heart volume shift during exertion. Hugo Knipping built a crank ergometer in 1928, combined it with a gas-metabolism apparatus in 1929, and called the result spiroergometry — which is, in all essentials, what a cardiopulmonary exercise test still is.

How a cardiologist turned it into a single number

Robert A. Bruce became the first chief of cardiology at the University of Washington in 1950, at thirty-four. He had been running patients on treadmills since 1949, watching the electrocardiogram change minute by minute as the work went up.

His problem was that the standard of the day, the Master two-step test, was fixed. The same steps for everybody — too punishing for a frail patient, far too easy for a fit one, and therefore useless as a common scale.

Bruce's solution, published in 1963 in — of all journals — Pediatrics, was to stage the test. Begin at 1.7 miles per hour on a 10% grade. Every three minutes, faster and steeper. Continue, if the subject can, to 6.0 miles per hour at 22%. One protocol that fits a cardiac patient and a trained athlete, because it does not ask how far you got. It asks how long you lasted.

That is the moment the treadmill became a scale. Your fitness turned into a time. A time converts into an estimate of oxygen uptake, and that estimate is the ancestor of the figure now sitting on your wrist.

The Seattle Heart Watch programme, begun in 1971 and running to more than ten thousand participants, established that the protocol was reproducible. It also acquired the limits every clinical test acquires — sensitivity around 70%, specificity around 80%, both varying widely. A stress test is a clinical procedure a doctor orders, supervises and interprets. It is not something to approximate at home.

How it got into your house

In 1968 Kenneth Cooper, a physician who had been working with the United States Air Force, published Aerobics, which gave ordinary people a scoring system for exercise.

One of its readers was William Staub, a mechanical engineer in New Jersey, who noticed that nothing on the market let a person do this at home at a sensible price, and built one himself. By the end of the decade the machine that had been a punishment, then a laboratory instrument, then a diagnostic device, was a consumer good for the spare room.

At every step the machine grows more voluntary and more private. At every step it keeps its original function: converting a person into a number.

Cubitt's number was revolutions. Smith's was cubic inches of air. Bruce's was minutes. Yours is an estimate of VO₂max — millilitres of oxygen per kilogram of body weight per minute — which your watch produces without measuring any oxygen whatsoever.

What the machine is actually evidence for

Two centuries of engineering deserve an honest accounting of what the resulting number is worth. The largest look at it is a 2018 analysis in JAMA Network Open: 122,007 adults who underwent treadmill testing at one American clinic between 1991 and 2014, mean age 53, followed for a median of 8.4 years.

Higher measured cardiorespiratory fitness was associated with lower all-cause mortality, and the finding that drew attention was the absence of a ceiling — no threshold beyond which the association flattened. Even people classified as "high" fitness carried more risk than those classified as "elite," and elite against least-fit gave a difference in mortality risk of around 80%.

Now the caveats, which are not decoration.

These were people referred for a stress test, which is not a random sample of anything. And an observational study of this kind cannot separate how much of the gap is produced by training from how much is produced by the underlying health that lets a person train at all. Fitness is partly something you do and partly something you have; this design cannot give you the ratio.

What survives is still substantial. Cardiorespiratory fitness, properly measured, is one of the most informative single figures anyone has found in a person — not because it is a lifestyle score, but because a great deal of biology has to be working for it to be high. It is associated with outcomes well beyond the heart, and among the first things to go when training stops.

What your wrist is doing instead

Your watch is not doing what Bruce's treadmill did. A laboratory test measures oxygen. Your watch models it — it observes heart rate against pace and infers what that relationship implies about capacity.

Sometimes the inference is good. A 2025 systematic review in Frontiers in Sports and Active Living pooled 13 validation studies covering 273 participants; seven of the thirteen showed acceptable validity for estimating VO₂max, with mean absolute percentage errors mostly in the 5–10% band. One came in above 15%.

Two findings there deserve more attention than the headline. First, the errors are not random: estimates ran high in less fit people and low in fitter ones, which means the number is pulled toward the middle of whatever population trained the algorithm. Second, optical sensors at the wrist performed worse than chest straps — no model recovers information the sensor never captured.

Beyond roughly 60 mL/kg/min the estimates become unreliable, and the review says plainly that these devices are not ready for elite endurance sport. It also notes what should be obvious and rarely is: the algorithms are proprietary and unauditable, and the validation samples are tiny — a median of about 21 people per study.

None of this makes the figure worthless. It makes it a trend rather than a measurement — which is what most wearable numbers actually are. These are wellness estimates, not medical measurements, and not a diagnosis of anything.

The bottom line

Cubitt built a machine to prove that work had been done. Two centuries later we buy the thing voluntarily and still want the same service from it: proof.

The evidence says the proof worth having is the direction, not the digit. The least reliable part of a wrist fitness estimate is its absolute value — the part people screenshot and compare. The reliable part is how it moves across two or three months on unchanged hardware and familiar terrain. Watch the line; ignore the number printed on it.

And if you want something you can observe rather than infer, notice how fast your heart rate comes down in the minute after you stop. It is directly measured rather than modelled, it needs no algorithm, and it tends to move in the same direction as the thing you were trying to estimate. Edward Smith would have approved. He counted breaths for a reason: they were the part he could actually see.