Gliding on ice is still an open physics question.

Most of us learned that the blade's pressure melts the ice, and the skater glides on the meltwater. It's a tidy story. It falls apart as soon as you put numbers on it. Ice is unusual: it is less dense than water, so squeezing it does lower its melting point. But only by about 0.0074 °C per bar (from the Clausius–Clapeyron relation).

The candidates, stated honestly

1. Premelting (a liquid-like surface layer). Even well below 0 °C, the top few molecular layers of ice are disordered and mobile. Faraday suggested a wet surface layer back in 1859. Modern measurements confirm a "quasi-liquid" layer whose thickness grows as you approach 0 °C (reviewed by Rosenberg in Physics Today, 2005).

2. Frictional heating. Bowden and Hughes argued in 1939 that the heat of sliding itself melts a thin film. It fits many observations (faster sliding often means lower friction), but direct measurements of the warming have been surprisingly elusive.

3. Mobile surface molecules. A 2018 study (Weber, Bonn and colleagues) linked friction to how mobile the surface molecules are. They found friction lowest at about −7 °C, the temperature used for speed-skating rinks. Warmer than that the ice softens and the slider digs in; colder, the surface is less mobile.

4. What the film is like. A 2019 experiment (Canale and colleagues) that probed the meltwater under a sliding probe found a film hundreds of nanometres thick that behaves far more viscously than ordinary water, more like oil.

5. Cold amorphisation. Molecular simulations published in 2025 (Atila, Sukhomlinov and Müser) suggest sliding scrambles the ice surface into a disordered, liquid-like layer without thermally melting it, and that the counter-surface's properties matter. They also note metal blades conduct heat over 20 times better than ice.

A skate blade isn't flat underneath. It's ground with a shallow concave groove along its length, the hollow, which leaves two sharp edges: the inside edge (towards your other foot) and the outside edge. Seen from the side, a figure blade is also gently curved (the "rocker"), so only part of it touches the ice at once, and the front carries toe picks for jumps and spins. Hockey blades have no toe picks.

Tilt onto an edge and it cuts a slight groove. The ice can now push sideways on the blade, which is what lets you turn, push and stop. Gliding on a flat blade, you'd have almost nothing to steer with.

The lean you need

To move on a circle of radius r at speed v, something must pull you toward the centre with a force of m × v² ÷ r. On ice, that's the sideways push of the edge. To balance, your body leans in so that the total force from the ice runs through your centre of mass: tan θ = v² ÷ (g × r).

Crossovers

In a forward crossover, the outside foot crosses over the inside one while both stay on edges: the outside skate on its inside edge, the inside skate on its outside edge. Each push is directed outward from the circle, and the ice's reaction pushes you inward and forward at once. It turns and accelerates you in one movement.

Try to push straight back on a blade and it just slides backwards along its length: there's almost no grip in that direction. Instead, a skater angles the pushing skate outward and drives it sideways on an edge. The ice pushes back perpendicular to the blade, and part of that reaction points forward. Speed skaters have refined this into a long, sideways stroke; beginners learn it as "swizzles" and "scooter pushes".

How tiny is ice friction?

Steel on good ice has a friction coefficient of roughly 0.005, depending on temperature and speed (Chapter 1). For our 70 kg skater, friction = 0.005 × 686 N ≈ 3.4 N, about the weight of a 350 g bottle of water.

How far can you coast?

With friction alone, the deceleration is just 3.4 ÷ 70 ≈ 0.05 m/s², and from 5 m/s you'd glide about 255 m. Add drag and the real coast falls to around 135 m (and much less on rough or soft ice). Either way, ice gives you an extraordinary amount of glide per push. That's why the first lesson is learning to stop (Chapter 6).

A spinning skater on a sharp blade feels almost no outside twisting force (torque): the ice and air exert only a little. When there's no outside torque, angular momentum stays constant: L = I × ω, where ω is the spin rate and I is the moment of inertia, roughly "how far your mass is from the spin axis", summed as mass × distance².

Why the arms matter so much

Distance counts squared. Treat each arm as roughly 4 kg concentrated 0.6 m from the axis: 4 × 0.6² ≈ 1.4 kg·m² per arm. Pulled in to 0.15 m it's 4 × 0.15² ≈ 0.09 kg·m². Two arms alone can change I by over 2.5 kg·m². (Real arms aren't point masses; the numbers show the scale, not a measurement.)

Why spins slow down, and why skaters get dizzy

A spin isn't perfectly free: the blade scrapes the ice and air drags on the limbs, so the spin slowly loses angular momentum. That's why skaters often finish by opening out. Dizziness comes from the fluid in the inner ear, which keeps swirling after you stop. Skaters reduce it with training, and by focusing their eyes as soon as they stop.

Once the blade leaves the ice, two things are fixed: the centre of mass follows a projectile path set by the take-off speed, and the angular momentum is whatever the skater generated on the ice (Chapter 4). In the air, the only way to rotate faster is to pull into a tighter position.

Airtime grows slowly with height

Rising to height h takes √(2h ÷ g), and falling back takes the same again.

The energy and the landing

Lifting a 50 kg skater's centre of mass by 0.5 m takes m × g × h ≈ 245 J, all of which comes back as speed on the way down (3.1 m/s at landing). Landing on one blade, the skater bends the knee and hip to stop that downward speed over some distance d. The average force is body weight × (1 + h ÷ d):

Peaks are higher than averages, which is why a soft knee, a strong landing position and progressive training matter so much. Coaches build up jumps over years for this reason.

Three ways to stop

Gliding along the blade there's almost no friction, so every stop works by turning a blade across the direction of travel so it scrapes. The snowplough pushes one or both blades out at an angle and shaves the ice gently. The T-stop drags the back skate at right angles behind the front one. The hockey stop swings both blades sideways at once, and friction plus shaving turns kinetic energy into heat and a spray of "snow".

Falling well

Everyone falls. The impact speed depends on how far your centre of mass drops: v = √(2gh). Standing tall, it's about 1 m (≈ 4.4 m/s at impact); crouched, about 0.5 m (≈ 3.1 m/s). Your head, falling from about 1.6 m, could reach ≈ 5.6 m/s. That's why coaches teach beginners to bend their knees, tuck their chin and go down sideways onto the hip and thigh, rather than throwing out a straight arm. Many rinks and coaches recommend helmets for beginners. Get up quickly from kneeling, because other skaters and blades are moving around you.

Rinks: ice made to order

Indoor rinks freeze a sheet of water over a slab cooled by chilled fluid running through pipes, and resurfacing machines periodically shave the top and lay down a thin film of fresh water that freezes smooth. Rinks tune their ice for use: harder, colder ice for speed and hockey, slightly softer for figure skating, where blades need to bite on take-offs and landings.

Natural ice: thickness is everything

Lakes and ponds freeze unevenly, and strength depends on the type of ice. The Minnesota Department of Natural Resources' general guidance for new, clear ice: under 10 cm (4 in), stay off; about 10 cm for activities on foot. White (snow) ice is about half as strong, so double the guidance. Ice can be thick in one spot and thin a few metres away. In India, natural skating happens at managed sites like Shimla's open-air rink and Ladakh's frozen ponds in deep winter. Skate only where the people managing the site say it's open.