CurlTi.me

CurlTi.me/Learn/Sweeping

What sweeping actually does

From the side of the sheet it looks like two people dragging a stone down the ice by force of effort. They are not. They are briefly changing the surface it is about to touch, and the whole effect fits inside a few percent.

Last updated 2026-08-16

The mechanism, in one paragraph#

A curling stone does not slide on ice. It slides on a very thin film of water that the stone's own pressure and motion create underneath its running band, and on the tips of the pebble sprayed onto the sheet. Brushing the ice a moment before the stone reaches it warms and polishes those tips. The stone then meets a slightly easier surface, so it travels further and — because the sideways bite that makes it curl comes from the same friction — curls less.

That is all of it. Nobody is dragging anything. The sweepers are changing the road a few centimetres ahead of the wheels.

The numbers

Full-effort sweeping multiplies friction by 0.934 — a 6.6% reduction. Sweeping one half of the path steers the stone about 71 mm away from the swept side over a full slide. Both are measured values, and both are the values the game's engine uses.

Sweeping only works on a slow stone#

This is the part that surprises people, and it is the part that explains everything sweepers actually do.

A brush head is 20 cm wide. A person sweeping hard manages about 4 strokes a second. Multiply those together and you get the fastest the brush can cover ground: above roughly 1.5 m/s, the strokes stop overlapping and there is simply untouched ice arriving under the stone between them. Sweeping a stone at that speed is exercise, not physics.

How much sweeping can do, against how fast the stone is going. Full effect below 0.5 metres per second, falling to nothing at 1.5, and nothing at all above that.nothinghalfall of it00.51.01.52.02.53.0a draw, crossing the hog lineboard weight, same pointa peel — nothing to be donestone speed (m/s)how much sweeping does
Full effect below 0.5 m/s, tailing off to nothing by 1.5 m/s. A draw crosses the far hog line at about 1.09 m/s, already inside the useful band; a peel crosses at nearly 3 m/s and can never be swept at all.

Since deceleration depends only on speed, every stone that comes to rest on the sheet spends its last 12.6 metres below 1.5 m/s — and its last 1.2 metres below 0.5 m/s, where sweeping is at full strength. That window is fixed. It is the same for a draw to the button and for a stone that only just crosses the hog line.

Which is why a takeout is barely sweepable: it never spends those 12.6 metres on the ice, because it has left the sheet first.

ShotSpeed crossing the far hog lineSpeed at the buttonWorth sweeping?
Draw to the button1.09 m/sstoppedYes — the whole way in.
Board weight takeout1.46 m/s1.02 m/sA little, and only near the house.
Peel weight2.91 m/s2.73 m/sNo. Nothing a brush can do.

Why friction rises as the stone slows#

The other reason the last few metres matter so much is that the ice gets harder work as the stone slows down. The coefficient of friction is not a constant; it follows

µ(v) = 0.00662 + 0.00231 / √v

so a stone at walking pace meets nearly twice the friction of one at delivery speed. This is the correction that makes real curling timings reproducible at all: with a constant coefficient you can match a correct hog-to-hog split or a correct total travel time, but never both at once.

Friction against a curling stone rises as the stone slows down, following mu of v equals 0.00662 plus 0.00231 over the square root of v.00.0060.0120.0180.51.01.52.02.53.0at releaseabout to stopstone speed (m/s)coefficient of friction
The friction a stone meets, against its speed. It rises steeply as the stone slows — which is why a stone that looks like it will be heavy so often is not.

It also explains a thing every curler notices and few can articulate: a stone that looks like it is running long usually is not, because the last few metres eat speed far faster than the first few did.

Sweeping one side#

Sweepers can brush only one half of the stone's path, and doing so steers it — away from the swept side. Brush the right-hand half and the stone drifts left. It is counter-intuitive until you see why: the swept half is easier ice, so the running band bites less on that side, and the stone rotates its way toward the side that still has grip.

Sweeping only one half of the stone's path steers it away from the swept side by about 71 millimetres over a full slide.Hog lineTee lineBack linebrushed on this side onlyso the stone drifts the other way
Brushing one half of the path. The effect is real, measurable and small: about 71 mm — under three inches — over a whole slide.

Three inches decides a lot of curling games. It does not decide any single badly thrown stone.

The rules of who may sweep what#

What good sweeping looks like#

Pressure, not speed. The brush has to be loaded with real downward force — serious sweepers put a substantial fraction of their body weight through the head — and the head has to stay in front of the stone rather than beside it. A fast, light stroke that never touches the ice hard enough is decoration.

The other half of the job is not sweeping. Sweepers are the only people who can see how the stone is actually running, and they call the weight back to the skip while it travels. A team that sweeps everything has thrown away its information channel and its ability to correct — which is why sweepers commit late, and why "clean" (brush lightly, touching nothing) is an instruction with a purpose.

Sweeping on CurlTi.me#

The game gives you the same lever the ice gives: sweep or do not, over any stretch of the stone's path, on one side or both. The engine uses the measured constants above rather than a feel that was tuned until sweeping seemed satisfying — so it will not rescue a badly thrown stone, and a peel cannot be swept at all, exactly as on the ice.