Transcript
The ice cave
In 1938, two scientists from Cambridge set up a laboratory inside a glacier.
They worked in a cave dug out of the ice above a research station at the Jungfraujoch, high in the Swiss Alps. Inside the cave, the temperature never rose above minus three degrees.
On the bench stood a hollow copper turntable. They could fill it with dry ice, or with liquid air, to make it colder still. On top, they froze a layer of ice, and planed it smooth with a small plane, like a carpenter shaving a board.
Then they set it spinning, with a tiny model ski riding on the ice. A fine thread ran from the ski to a flat spring. A beam of light, bounced off a little mirror, showed how hard the ice pulled back.
Outside, on a level patch of snow, they lashed two wooden skis together and loaded them with rocks. Between ten in the morning and noon, in a light mist, they dragged them along with a spring balance.
The snow, they noted, had fallen two days earlier, just outside the laboratory.
They were trying to answer a question that sounds as if it should have been settled long before.
Why does anything slide on ice at all?
This is Slidensa — stories of ice, gravity and speed.
Ask almost anyone why ice is slippery, and you will probably hear the same answer.
The weight on a skate blade presses down on a tiny area. That pressure melts the ice. And the skater glides on a thin layer of water.
It is neat. It is memorable. And as an explanation of skating, skiing or sliding, it is mostly wrong.
Not completely wrong. Pressure really does lower the melting point of ice. But the effect is far too small to explain what happens under a blade, or a runner, especially in the cold.
So that is the question for this episode.
Why is ice slippery? And what does the answer mean for a sled, running on steel at well over a hundred kilometres an hour?
Here is the honest answer, up front. There is no single one. Physicists are still arguing about it. And while physicists argue, the sliding sports build rules around the parts they can measure.
The answer everyone learns
The pressure explanation begins with two brothers.
William Thomson, later Lord Kelvin, was a young professor in Glasgow. He posed a puzzle to his older brother James, an engineer.
Water expands when it freezes. So, in principle, freezing water can push things. Could you turn water into ice by a purely mechanical process, and use it to do work, over and over, for nothing?
That would be a perpetual motion machine. And James knew that could not be right.
His way out was this. If pressure lowers the temperature at which water freezes, the loophole closes.
On the second of January 1849, his prediction was read to the Royal Society of Edinburgh. One extra atmosphere of pressure, he calculated, would lower the freezing point of water by seventy-five ten-thousandths of a degree.
That is tiny. So tiny, he wrote, that it explained why nobody had ever noticed it.
His brother set out to measure it anyway. William built a thermometer filled with ether, so sensitive that its column rose more than two inches for a single degree Fahrenheit. He sealed it in glass, packed it in ice and water inside a pressure vessel, and squeezed.
On the tenth of January 1850, he wrote to Michael Faraday in London. “I have today succeeded.”
Eight atmospheres lowered the freezing point by about a tenth of a degree Fahrenheit. His results matched his brother’s prediction almost perfectly. So perfectly that William admitted the close agreement was partly down to chance.
Neither brother said a word about skating. James later used the idea to think about glaciers, and how solid ice can slowly flow.
That step came in 1886, from a young engineer in Dublin named John Joly.
A skate blade is curved along its length, Joly pointed out. On smooth ice it first touches along a line of, as he put it, indefinite thinness. So the pressure is enormous.
He worked through an example. A skater weighing a hundred and forty pounds, standing on a sliver of blade, would press with seven thousand pounds on every square inch. That, he wrote, was enough to melt the ice at minus three and a half degrees.
The skater, in fact, glides about on a narrow film of water.
Joly even noticed a problem. On very cold ice, he wrote, skaters find it difficult to get any bite. He read that as a matter of grip, and to cope with it he recommended skates ground hollow, to bite better.
He ended with a flourish. In the expression “as slippery as ice”, he wrote, there is revealed a consensus of opinion about the strange nature of ice.
In 1899, the Manchester engineer Osborne Reynolds made the same argument. And pressure melting became the answer that most people would learn.
Do the sums
The trouble starts when you do the arithmetic.
To lower the melting point of ice by one degree, you need about a hundred and thirty atmospheres of pressure.
To melt ice at minus ten, you need more than a thousand. At minus twenty, about nineteen hundred.
And below about minus twenty-two, pressure alone cannot melt ordinary ice along that path at all. Squeezed that hard, it turns into a different kind of solid ice instead.
Yet people skate in much colder weather than that.
In 1995, the American snow scientist Samuel Colbeck opened his summary bluntly. Pressure melting, he wrote, cannot be responsible for the low friction of ice.
Bowden and Hughes, in their ice cave, did this sum before they did anything else.
A skier weighing seventy-five kilograms, they worked out, spreads that weight over the whole surface of the skis. That pressure could only melt snow warmer than about one ten-thousandth of a degree below zero.
Suppose the skis really touched the snow on only a thousandth of their surface. The limit would still be about a tenth of a degree.
To melt snow at minus twenty, the real contact would have to be less than a hundred-thousandth of the ski. And yet, they wrote, experience shows that skis slide quite readily in exactly those conditions.
There is one honest complication. When steel meets ice, it does not touch everywhere. It touches at tiny high points, and there the local pressure can be very high. But ice is not infinitely strong. Under that kind of pressure it can also crush, or yield, or be ploughed aside.
So pressure may matter at a few points, close to zero. As a general explanation, it fails.
Bowden and Hughes also measured what happens as snow gets colder. Their waxed wooden skis needed a pull of about four percent of their load at zero. At about minus ten, it was eighteen percent. At minus forty, it was forty percent.
On their turntable, they cooled the ice much further, to a hundred and forty degrees below zero. The friction there was five or six times what it was near zero.
And there is evidence from the cold itself.
Very cold snow is not slippery at all. On the nineteenth of February 1912, Robert Falcon Scott and his companions were hauling their sledge back from the South Pole. The temperature was about minus twenty-seven.
The surface, Scott wrote in his journal, was like pulling over desert sand. Not the least glide in the world.
He and his companions died on that march, a few weeks later.
Charles Wright, the physicist who had stayed at the expedition’s base, later summed up what the sledgers had found. Below about minus eighteen, he wrote, the friction seemed to increase steadily as the temperature fell.
No single number
To talk about slipperiness, physicists use a friction coefficient. Put simply, it is how hard you have to pull something to keep it sliding, as a share of its weight.
For ice, that number refuses to stay still.
In 2018, a team in Amsterdam slid a small steel ball very slowly across ice, at temperatures from minus a hundred up to zero.
At minus a hundred, the friction was about half the ball’s weight. That is about what you would get from steel on glass. Ice that cold is not slippery.
As the ice warmed, the friction fell, to about fifty times lower near minus seven. Then, closer to zero, it rose again. The ice had become soft enough for the ball to plough a groove in it.
Real skaters give a similar picture. In 1992, Dutch researchers fitted speed skaters with instrumented skates. The friction averaged about half a percent of their weight. It was lowest when the ice surface was near minus seven.
Several tests have landed near that figure. But each one is true under its own conditions: its own shape of blade, its own speed, its own load. Change those, and the lowest point moves. It is not a universal best temperature, and it is not a number for a sled track.
What about sleds?
In a controlled test published in 2011, a set of runners from a two-man bob gave a friction coefficient of about four thousandths. A broader radar analysis of bobs on the ice gave about five thousandths.
In a laboratory in Germany, a steel pin was dragged across ice at up to two hundred and forty kilometres an hour. Its friction never fell below eight thousandths.
Even the effect of speed depends on the conditions. In some experiments, friction falls as you go faster. In others, it rises. Real speed skaters, in that 1992 study, found it rising with speed.
And in 2018, Latvian researchers ran a loaded skeleton sled down a short ice track, again and again, in different weather. It ran fastest when the air was at about zero.
And over a whole bobsleigh run, with bumps, curves and worn ice, one engineering estimate puts the average nearer fifteen thousandths.
Those are very small numbers. For a four-man bob at its weight limit, crew included, five thousandths means a resisting force of about the weight of three kilograms.
So, how slippery is ice? It depends on the temperature, the speed, the load, the shape of the runner, its roughness, and how worn the ice is.
A game has to choose.
The current Slidensa simulation gives ordinary runner friction one fixed number. The bobs and the skeleton use six thousandths. The luge uses four thousandths.
That number never changes with temperature, speed, the place on the track, or the load in a curve. Skidding and steering losses are worked out separately.
Those values were not measured from ice. They were set together with drag, starts and handling, so that the sleds behave believably against one another.
Real ice is not so obliging. And to see why, we have to go back to 1850, and to the man William Thomson had written to.
Faraday’s surface
It was the evening of the seventh of June 1850. In London, Michael Faraday gave a lecture at the Royal Institution.
Two pieces of moist ice, he told his audience, will consolidate into one. That is why damp snow can be pressed into a snowball, and hard-frozen snow cannot.
In the same lecture, he mentioned the new result from Glasgow. Mr Thomson, he said, has shown that immense pressure will prevent water from freezing. The two ideas shared a stage from the very start.
Faraday’s own explanation was subtle. A particle of water touching ice on several sides, he suggested, freezes more readily than one touching ice on only one side.
Turn that around, and it says something about the ice itself. Its outermost layer, with solid ice beneath and nothing above, is not quite as firmly frozen as the inside.
James Thomson disagreed. Where two pieces of ice touch, he argued, the pressure melts them, and the water freezes again when the pressure is released. The physicist John Tyndall gave the effect a name: regelation, or refreezing.
In 1860, Faraday tried to settle it. He moored blocks of ice under ice-cold water, on threads weighted with lead, so that if they were pushed together, they sprang apart.
Then he let them touch, very slightly.
They stuck. They stayed stuck, he reported, despite the force pulling them apart. They held for twenty-four hours or more. In another version, twisted against each other, they did not move in thirty-six hours.
No force was pressing the blocks together. They were being pulled apart. Thomson replied again, and did not give way. And for almost a century, the pressure explanation mostly won.
One caution here. Faraday was explaining why ice sticks to ice. He never talked about skating. Regelation shows that ice behaves unusually where it meets water or air. It does not, by itself, explain why ice is slippery.
But his idea about the outer layer turned out to be right in spirit.
Measurements over the past fifty years have found that, even well below zero, the surface of ice carries a thin, disordered, liquid-like layer. In the cold, it is just a few molecules thick. Closer to zero, it grows.
Scientists call this premelting.
It seems to grow in steps. In 2017, one study found the outer layer of molecules already disordered below about minus thirty-eight. The next layer gave way quite suddenly, at about minus sixteen.
Under an optical microscope, patches of something that looks like real liquid appear only within a degree or two of zero, as droplets and thin films.
It is not ordinary water. And it has been hard to measure. Early results from different instruments disagreed by as much as a hundred times. Better methods have narrowed the range since.
In 2024, a team in China photographed an ice surface almost atom by atom. The disorder began at tiny defects, at about a hundred and fifty degrees below zero. Ice, one of the researchers told Science News, is not so perfect.
And the surface can mend itself. In the Amsterdam experiments, a scratch made with a screwdriver at minus ten had almost completely healed seven minutes later.
There is one more caution. All this describes the bare ice, open to the air. Ice under a steel runner is squeezed, sheared and heated. It is not the same situation.
Which brings us back to the cave.
Heat from sliding
Frank Philip Bowden was a chemist from Tasmania, working in Cambridge, and an excellent skier. According to a colleague, he was once snowbound in the Alps, and began to wonder why the friction of skis on snow is usually so low.
In 1938 he joined a glaciological expedition to the Jungfraujoch. With his colleague Hughes, he went looking for the answer. The Ski Club of Great Britain helped to pay for the work.
They were not alone in the ice. On the same expedition, the young chemist Max Perutz worked in a chamber cut into the glacier. He would later win a Nobel Prize. His cold laboratory held at about minus four all summer.
Bowden and Hughes came up with a clever test.
Take two model skis, the same size and shape. Make one of brass, a metal that conducts heat very well. Make the other of ebonite, a hard black rubber that hardly conducts heat at all.
Now think about what each theory predicts.
If the water under a ski comes from pressure melting, the melting needs heat from somewhere. Brass can bring heat in from around it. On that theory, brass should slide better.
But if the heat comes from friction itself, made right at the contact, then brass will carry that heat away. Less ice melts. On this theory, the brass should slide worse.
Near zero, the two skis slid the same. On colder ice, the friction of the brass was considerably greater.
They tried it another way. A hollow ski with a copper sole slid worse when they filled it with mercury, which carried the heat away even faster.
The explorer Fridtjof Nansen had noticed the same thing in the Arctic, in 1895. On his expedition, sledges with nickel-plated runners were at least half as hard again to pull as sledges on tarred maple.
Bowden and Hughes even looked for the water electrically. They froze their ice from slightly salty water, and set electrodes into the ebonite ski. On cold ice, the current told them that the melting happened only locally, at isolated points of contact. Close to zero, the film joined up. It was, they wrote, still invisible.
There was one more clue. Starting a slider moving took far more force than keeping it going. At minus twelve, they measured about ten times more to start it than to keep it sliding. Heat from friction needs movement. Once you stop, the help is gone.
Their verdict was careful. Pressure melting may play some part, they wrote. But the low friction of fast-moving things on ice is due in a large measure to local melting by frictional heating.
They were not the first to suspect it. In the winter of 1912, in a hut in Antarctica, the physicist on Scott’s expedition, Charles Wright, gave a lecture about sledge runners. The runners, he said, probably melt the millions of crystal points they glide over.
The sledge is running upon water.
It was an informed guess, not an experiment. Bowden and Hughes put numbers to it.
Later work backed them up. In 1953, Bowden reported that a new plastic, PTFE, better known by the trade name Teflon, gave very low friction on snow and ice under all conditions. A colleague noted that the paper caught the eye of ski and toboggan enthusiasts.
In 1976, three researchers in Bristol hung a pendulum against a spinning drum of ice and refined the heating theory with numbers. Most of the heat, they found, does not go into melting at all. It flows away, into the slider and into the ice.
And in 1997, American researchers fixed tiny thermometers to skates. Sliding, they found, leaves a warm track on the ice behind the blade.
But heat cannot be the whole story either. As the chemist Robert Rosenberg argued in 2005, neither theory explains why ice is slippery before any sliding begins.
The argument today
That might sound like the end of the story. It is not.
In 2018, Daniel Bonn’s group in Amsterdam, with colleagues in Germany, ran the slow steel-ball experiment you heard about earlier. In their computer models, the water molecules at the surface of ice were loosely held, and rolled across it. They tied the low friction to those mobile molecules. In their slow tests, frictional heating did not matter much.
In 2019, a team in Paris measured the film under a small bead sliding on ice. It was thicker than the resting surface layer. But it was not runny. It was up to a hundred times more viscous than water, as viscous as an oil. They suggested it might be a slurry of water and tiny fragments of ice.
That also suggested a fresh reason why wax works. Water-repelling coatings on the bead cut the friction sharply.
In 2022, computer simulations by a group in Madrid found premelting, pressure and frictional heating all working together. The film they saw was about a millionth of a millimetre thick.
A simulation follow-up found that this film became far runnier when it was sheared fast. That does not sit easily with the thick, oily film measured in Paris. The two groups were looking at very different conditions.
In 2025, simulations from a group in Germany went further. Under their model slider, the ice did not have to melt in the usual sense. Sliding scrambled it into a disordered, glassy layer. That happened fastest at the lowest temperatures.
It is a simulation, not proof. But it offers a new explanation for poor glide in deep cold: the layer still forms, but it is extremely viscous. The university’s press release compared it with something more viscous than honey.
In 2026, another team posted a study that has not yet been reviewed by other scientists. In their model, heat from sliding is the key, without the ice fully melting. Below a certain speed, about ten centimetres a second, a foot keeps its grip. Above it, a quick scuff ends, as they put it, in catastrophe.
Researchers at the United States Army’s cold-regions laboratory weighed five rival explanations: pressure melting, melting by friction, the premelted layer, abrasion, and slushy mixtures of ice and water. They pointed to field measurements in which the average temperature at the contact stayed below zero. Their own case stressed brittle ice, scraping and wear. That does not rule out tiny hot spots. But it does not fit one smooth, even film of water.
In 2025 the science magazine Quanta asked the researchers what they made of all this.
Bonn’s objection to the heating explanation was simple. You only melt the ice behind you, he said, not the ice you are actually skating on.
Luis MacDowell, in Madrid, thought the rival ideas all operate at the same time, to one degree or another.
And Bonn added that ice researchers do have different and contradictory opinions. But they don’t really tell each other that they disagree.
So, more than a hundred and seventy years after the Thomson brothers, the question is still open.
Under a runner
The people who race on ice have not waited for the physicists.
Start with the runners. In bobsleigh and skeleton, the international federation allows one kind of steel. It comes from a single designated factory. The rules say why: to keep the competition fair, to control costs, and to ban exotic materials and treatments.
Any treatment that changes the steel, even locally, is forbidden. So is any plating or coating. Runners may be polished only with abrasives the federation supplies, and the grade of sandpaper is fixed before each season.
Within those limits, teams can shape, sand and polish. They may add nothing.
In skeleton, the testing goes further. Before each run of the day, the testing protocol calls for the runners to be wiped with an official cleaning fluid.
Luge still does not prescribe one standard steel. Its current rules control the shape of the steels instead. In 2024, the federation voted to standardise the dimensions of sled parts, in stages, from the 2026–27 season. In 2018, members of the American women’s bobsleigh team described spending three to four hours on each set of runners. They started at a coarse grit and worked down, grit by grit. Between races, the runners travelled in padded gun cases.
How much does ice drag matter, compared with air? One British study modelled a women’s skeleton run at Winterberg, in Germany, in 2015.
In that model, about sixty percent of the lost energy was classed as air drag, and about forty percent as ice drag. The second share included runner friction and any knocks against the ice. For heavier riders, the model put ice drag as high as sixty percent. The authors expected it to dominate in the curves, with air drag taking over on the straights.
If all the heat from runner friction stayed in the steel, the same study calculated, the runners would warm by about fifteen degrees. Yet the authors noted that runners finish cold, and often covered in ice. The heat does not stay in the runners.
Track crews face a different problem: not finding one perfect temperature, but making ice that stays consistent. Even the experts disagree on where the window sits. Some say colder ice is always faster. One Canadian model predicted its highest speeds between minus ten and minus twenty. Its author said that agreed with observations at the Calgary track.
At the 2018 Winter Olympics in Pyeongchang, the head of the track’s ice crew, Markus Aschauer, aimed for ice at about minus two or three degrees.
If the ice is too warm, he explained, it starts to frost or even melt. But if it is too cold, it becomes very hard, like glass, and starts to crack.
At Park City, in the United States, the track’s plant manager described the daily balancing act. It is a fine line, she said, between what the sun is doing to the track and what the humidity level is.
Germany’s institute for sports equipment puts it more bluntly. Ice close to its melting point, its engineers wrote, makes horrible conditions for the sport. It is slow, and it wears out during a race.
And the sun is part of the race. Shades over the curves must stay open during training and competition, unless the ice is at real risk. Everyone faces the same sky.
Warm runners and the rules
Back in 1938, Bowden and Hughes tried one more thing. They warmed their model skis to ten degrees, and slid them on ice at zero.
This time, the brass ski won. Under the heavier load, its friction was about five thousandths, lower than the ebonite.
Why would a warm runner help? The engineers who study sled runners describe the lubricating water film as coming from the premelted surface and from frictional heat. A cold runner draws heat away from the contact. A warmer one, it seems likely, draws less.
Germany’s sports equipment institute is clear about the result. In the sliding sports, its engineers wrote, the runner’s temperature has a high effect on the finish time.
But warmth is not free speed without limit. In the German high-speed tests, on ice at minus six, the friction rose again at the highest speeds, as water built up on the surface. And no published result says how much time one degree of runner warmth is worth.
The rulebooks do not wait for that number either.
In bobsleigh, a reference runner of the same standard steel is hung up in the open air forty-five minutes before the race. When it goes out, it must read between eighteen and twenty-two degrees. Then it cools, in the same weather as the sleds.
Officials press an electronic thermometer against the side of each runner. A sled’s runners may be at most four degrees above the reference. If a runner is over the limit, it is measured again straight away. Over the limit twice, and the sled is disqualified.
In bobsleigh, the checks may be made before each heat, as the officials decide. At the start, a board shows the reference runner’s temperature, the air and the ice, updated through the race.
By then, all work on the sleds must be finished. The sleds must lie upside down, uncovered, with their runners facing the sky. Cooling the runners with snow or ice is not allowed. In skeleton, the sled is carried with its runners away from the athlete’s body. And in the waiting area, skeleton athletes may not even wear gloves.
Luge does it differently. A control blade is fixed in the shade at the start, out of the wind, at least four hours before the race. Each sled’s steels are measured on their running surface, to a tenth of a degree, just before the start. They may be no more than five degrees warmer than the control.
The numbers go on a board that everyone can see. A third measuring device waits at the finish, for control checks. At championships and the Olympics, the best ten have their steels checked after the second and fourth runs. Even the forerunners, who test the track before a race, are checked. If their steels are too warm, they cannot run.
Rules like these are written for a reason. At the 1994 Winter Olympics in Lillehammer, an American four-man bob was disqualified because its runners were warmer than the rules allowed.
There is an earlier Olympic scandal about warm runners, too. That is a story for a coming episode.
There is one place where none of this can happen. A warm runner is one thing the current Slidensa simulation cannot represent. Its runners have no temperature at all.
The question in the cave
So, why is ice slippery?
Not mainly because pressure melts it. The numbers rule that out in the cold.
Partly because the surface of ice is never quite as solid as the inside, even far below zero. Partly because sliding heats it, shears it, and perhaps scrambles it into something that is neither ice nor water. And partly for reasons that are being argued about right now.
Which matters most depends on how cold the ice is, how fast you are going, how hard you press, and what you are sliding on.
The Thomson brothers were partly right. So was Faraday. So were Bowden and Hughes. None of them had the whole answer, and nobody does yet.
And the next experiment may change the picture again.
But the sliders do not need the whole answer. They need the steel to be the same for everyone. They need a thermometer on the runners, a board at the start, and an ice crew watching the sun.
Go back, one last time, to the plateau above the Jungfraujoch, in 1938. Two wooden skis, lashed together and loaded with rocks, pulled through the mist over snow that fell two days before.
The question they were asking is still open.
Sources, a transcript and the project are at slidensa.com.
Slidensa is an independent game about bobsleigh, luge and skeleton.