How much does release speed change where a curling stone stops?

Enormously, relative to how small the change is. Across the whole useful range of draw weight — the front of the house to the back of it — the release speed varies by only about seven centimetres per second. Two centimetres per second is roughly a foot of draw.

How much difference does ice speed make in curling?

Ten per cent more friction costs about eight and a half feet of draw. Friction and draw distance are exact reciprocals: scale friction by k and the stone travels 1/k as far. That is why the same delivery finishes short of the house on club ice and past the back line on championship ice.

How much further does sweeping make a curling stone travel?

Full-effort brushing multiplies friction by about 0.934, a reduction of 6.6%, but it only works below about 1.5 metres per second and is fully effective only below 0.5 — so sweeping earns its distance late, in the last stretch, which is why sweepers commit when they do.

CurlTi.me/Learn/Draw simulator

The draw simulator

The one thing about curling that prose cannot teach: how tiny the difference between a guard and a takeout is. Two centimetres per second is a foot of draw. Drag it and see.

Last updated 2026-08-18

The simulator#

Release speed 2.194 m/s

The speed the stone crosses the near hog line at.

Ice friction ×1.00 (calibration)

Below 1.00 is keen ice. Above is heavy.

Stops at 126.0 ft from the hack.
Hog to hog: 13.31 s
On the button

The stone marks where the current settings leave it. Distances are measured from the delivering hack, which is the convention the rest of this site uses.

With scripting turned off the sliders do nothing and the table further down carries the same information. Nothing here is hidden behind the controls.

Four things worth trying#

How small the useful range is#

Set the ice to ×1.00 and move the speed slider between 2.126 and 2.194 m/s. That is the difference between the top of the house and the button — a called 4 and a called 7 — and it is seven centimetres per second. The entire vocabulary of draw weight lives inside a window narrower than a brisk walk changes by.

This is the real answer to why curling is hard, and why the sport spends so much effort on a repeatable slide. See the delivery.

Why your first stone of the night is always wrong#

Leave the speed at 2.194 m/s — a perfect draw to the button on the calibration sheet — and move the ice slider instead.

Ice	Friction	That same delivery stops at
Club	×1.10	117.5 ft
Normal	×1.00	126.0 ft
Championship	×0.85	142.4 ft

Nearly twenty-five feet between the slowest and the fastest, from a delivery that felt identical in the hand. That is twice the width of the house, and it is why reading the ice is most of the skill.

Why sweepers commit late#

Set a shot that finishes just short — say 2.126 m/s on ×1.05 ice — and turn sweeping on. The stone gains a few feet.

Now do the same thing at takeout speed, somewhere above 3 m/s, and turn sweeping on again. It gains almost nothing.

That difference is not an artefact. Sweeping is fully effective below 0.5 m/s, useless above 1.5 m/s, and tapers between. A brush head is only 20 cm wide and a person tops out around 4 strokes a second, so past a certain speed the strokes stop covering the ice the stone is about to reach. A hard takeout is barely sweepable at all.

The stopwatch runs backwards#

This one needs doing carefully, because the obvious version of the experiment shows the opposite of the real effect.

Drag the ice slider on its own, leaving the speed alone, and the hog-to-hog time goes down as the ice gets keener — of course it does, the stone is being slowed less and it is going further. That is not what curlers mean by fast ice and a slow watch.

The real comparison holds the finishing position fixed instead. Draw to the button on each sheet — change the ice, then move the speed slider until it reads 126 ft again — and read the split:

Sheet	Release speed for the button	Hog-to-hog split
Club (×1.10, heavy)	2.295 m/s	12.72 s
Normal (×1.00, the calibration sheet)	2.194 m/s	13.31 s
Championship (×0.85, keen)	2.032 m/s	14.37 s

Keener ice needs a gentler delivery to finish in the same place, so the stone spends longer crossing the 72 feet between the lines. Fast ice, slow watch. This catches everybody, and it is the single most useful thing to have straight before you start timing stones — a team that reads a long split as a heavy delivery will correct in exactly the wrong direction.

What it is actually computing#

Not an animation of a pre-computed answer. Every time you move a slider the page integrates the slide, in half-millisecond steps, using the same model the game engine uses.

The friction law:

μ(v) = 0.00662 + 0.00231 / √v

Friction rises as the stone slows, which is the correction that makes real curling timings reproducible — a constant coefficient cannot produce a correct hog-to-hog split and a correct total distance at the same time.

Two other pieces of the engine come with it:

The carry. Friction starts at the hog line, not at the hack, because a curler slides out of the hack holding the stone and releases it there. Model it as a push from the hack instead and a draw to the button needs 2.54 m/s rather than 2.19, and takes four seconds longer.
Speed-dependent sweeping. Full-effort brushing multiplies friction by 0.934, tapered by how fast the stone is going, exactly as described above.

Why this is worth trusting

The constants in the browser copy are asserted against the engine's own constants by apps/web/test/learn.test.ts, so a change to the physics fails the test rather than quietly leaving one simulator on this site disagreeing with the other. Set the speed to a value from the weight table below and the answer matches it to a tenth of a foot.

The same thing as a table#

The numbered weight scale, on the calibration sheet. Set the slider to any of these speeds and the simulator will land on the same answer.

Called	Where it should finish	Release speed	Stops at	Hog to hog
1	a deep guard	2.019 m/s	111 ft	16.48 s
2	a guard	2.056 m/s	114 ft	15.58 s
3	a tight guard	2.091 m/s	117 ft	14.87 s
4	the top of the house	2.126 m/s	120 ft	14.27 s
5	the top of the eight-foot	2.149 m/s	122 ft	13.92 s
6	the top of the four-foot	2.172 m/s	124 ft	13.60 s
7	the button	2.194 m/s	126 ft	13.31 s
8	the back of the four-foot	2.216 m/s	128 ft	13.04 s
9	the back of the eight-foot	2.238 m/s	130 ft	12.78 s
10	the back line	2.26 m/s	132 ft	12.55 s

Throw one for real The weight scale, explained

Sources#

The curl mysteryWhy this simulator models the curl the way it does, and why that is an engineering choice rather than a claim about nature.
Ice speed, and what to expect from a sheetThe friction-and-draw reciprocal, the break-in effect, and how to read a sheet with a stopwatch.
What sweeping doesThe published measurements the sweep model is built from.