Do you push the stone in curling?

No, and this is the most commonly misunderstood thing about the sport. The curler slides out of the hack carrying the stone at a near-constant speed for the 33 feet to the hog line, and releases it there. The stone only begins to slow once it has been let go. A stone shoved from the hack would need a much higher release speed and would take about four seconds longer to reach the far end.

When do you put the turn on a curling stone?

Late — over roughly the last two or three feet of the slide. The handle starts at 10 o’clock for an in-turn or 2 o’clock for an out-turn and is rotated to 12 o’clock as you release. Turning early makes the rotation rate inconsistent, and rotation rate is one of the few things about a delivery you want to be boring.

How many rotations should a curling stone have?

About two and a half to three and a half over the length of the sheet — roughly one rotation every four or five seconds. More than that does not make the stone curl further, because curl is essentially flat in rotation rate. A stone barely turning at all is the problem case: it curls unpredictably.

Is toe sliding bad for your knees?

A published pilot biomechanical study found toe sliding produces a larger moment arm at the knee than flat-foot sliding, which is the mechanical basis for the advice most coaches already gave. Separately, 79% of competitive curlers report curling-related musculoskeletal pain, most commonly in the knee. Neither finding says never toe-slide; both say the load is real and worth managing.

CurlTi.me/Learn/The delivery

The delivery

You cannot learn this from a page. What a page can do is tell you the four things that are counter-intuitive, so the first hour on the ice makes sense instead of just hurting.

Last updated 2026-08-18

Read this first

Nobody has ever learned to deliver a curling stone by reading. Book a learn-to-curl session — every club runs them, they cost very little and the equipment is provided. This page is for understanding what you are being asked to do, which makes that session about twice as useful.

The carry: the thing everybody gets wrong#

A curling delivery is not a push. The curler slides out of the hack holding the stone, travelling with it at a nearly constant speed for the thirty-three feet to the hog line, and lets go there. Only after release does the stone start to slow down.

Two regimes, one delivery. For 33 feet the stone is being carried and nothing is slowing it. Everything the rest of this site says about weight and timing happens after the release.

This is not a technicality. It changes every number:

A stone crosses the near hog line at exactly its release speed, which is why a stopwatch reading between the two hog lines measures the slide cleanly and is the standard timing convention.
A back-line-to-hog time measures the carry instead — about 3.7 seconds at draw weight — which is why coaches use it to check a curler's slide rather than the ice.
If you model the stone as pushed from the hack, a draw to the button needs a release speed of about 2.54 m/s instead of the measured 2.19, and the shot takes four seconds longer. The engine behind this site models the carry for exactly this reason.

The hack#

Two rubber footholds sit either side of the centre line at each end. A right-handed curler uses the one on the left of the centre line, a left-hander the one on the right — that is a rule, not a convention, and a stone delivered from the wrong hack is removed from play.

What happens in the hack decides most of the shot:

The foot points at the target. The ball of the hack foot sits high in the hack and the toe points down the line the skip has given. The foot sets the line of the whole delivery, and an error there is angular — one degree out at the hack is about two feet out by the time the stone reaches the far button. Nothing later in the delivery can recover it.
The stone is set before you move. Handle at 10 o'clock or 2 o'clock, stone on the line of delivery, weight settled.
Hips back and low. Lower than the head. This is what lets the push be a leg movement rather than a lunge.

The slide, and balance#

Push out of the hack with the leg, bring the sliding foot under the chest, and ride. The single technical goal of the whole slide is that nothing changes: no steering, no leaning, no adjusting.

The classic diagnostic, and the fairest one:

Take a practice slide out of the hack with no stone and no broom. If you can do it in balance, you are balanced. If you cannot, you have been leaning on something.

That "something" is usually the broom, and it is the most common fault in curling. A broom held across the body is a stabiliser, not a crutch: once the slide is established there should be no downward pressure on it, and none on the stone either. A curler who is pressing down on the stone is steering it, and a curler who is pressing on the broom will fall over the moment the broom catches.

Look at the skip's brush, not at the stone. The stone will go where you were looking.

Phase	What matters
Line	Hack foot pointed down the line the skip has given you. Everything else follows the foot; if the foot is wrong, nothing later can fix it.
Stone	In front of you, on the line, handle at 10 o’clock for an in-turn or 2 o’clock for an out-turn — set before you move, not during.
Balance	Weight forward, hips lower than head, no downward pressure on the stone or the broom.
Push	Out of the hack with the leg. The stone does not get pushed; it gets carried.
Slide	Sliding foot under the chest, trailing leg relaxed behind. Look at the broom, not at the stone.
Turn	Rotate the handle to 12 o’clock over the last two or three feet, and let go. Roughly three rotations over the sheet.

The turn#

The handle starts at 10 o'clock for an in-turn or 2 o'clock for an out-turn, and is rotated to 12 o'clock over the last two or three feet of the slide as the stone is released. Late, gently, and the same way every time.

Aim for about two and a half to three and a half rotations over the length of the sheet — roughly one turn every four or five seconds, which looks much slower than people expect. Two things to know about why:

More rotation does not mean more curl. Across the normal range total curl is essentially flat in rotation rate, so spinning it harder buys you nothing at all. The physics of that is on why a curling stone curls.
Too little rotation is the real fault. A stone barely turning curls weakly and unpredictably. Murata's measurements describe the minimum-rotation case as behaving like a knuckleball, and recommend a slight, deliberate rotation precisely to avoid it.

Where weight actually comes from#

Not from the arm. Weight comes from the speed of the slide, which comes from the leg drive out of the hack, and the arm's job is to let go without changing anything.

2.19 m/srelease speed for a draw to the button, on calibration ice
0.07 m/sroughly, between the top of the house and the back of it
13.3 shog-to-hog split for that draw
3.7 sback line to hog line — the carry

Look at the second of those. The entire useful range of draw weight — front of the house to the back of it — is a few centimetres per second of release speed. That is the precision the delivery is trying to repeat, and it is why the answer to almost every weight problem is "make the slide more consistent" rather than "push harder".

It is also why sweeping exists: a stone that leaves the hand 2% quick is a stone the sweepers can absorb. See sweeping and calling weight by number.

Flat-foot and toe sliding#

Two schools, distinguished by what the trailing leg does.

	Flat-foot slide	Toe slide
Trailing foot	Sole flat, dragging behind	Rolled onto the top of the toe
Feel	More stable, easier to learn	Lower and longer; favoured by many elite curlers
Knee load	Smaller moment arm at the knee	Larger moment arm at the knee, in the published measurements
Best for	Almost everybody starting out, and most curlers permanently	Players with the mobility for it who want the lower position

There is no correct answer, but there is a sensible default: start flat-footed. It is easier to balance, it is kinder to the knee, and the number of games lost to an unstylish delivery is zero.

What this does to your body#

Curling is gentle in the way that gardening is gentle: nobody gets hit, and plenty of people hurt. The research is small but consistent.

79% of competitive curlers in one US survey reported curling-related musculoskeletal pain, most often the knee (54%), then the back (33%) and the shoulder (20%). Sweeping and delivering were the activities most likely to provoke it.
Time-loss injuries were estimated at about two per thousand athlete exposures, which is low. This is a sport of accumulated load rather than acute injury.
A pilot biomechanical study found toe sliding produces a larger moment arm at the knee than flat-foot sliding.

The practical reading: warm up before you throw, do not learn on a cold knee, and if the crouch is the thing hurting you, a delivery stick is not a defeat. It is how a great many people are still curling in their eighties.

Sources#

Primary Self-reported injury patterns among competitive curlers in the United States British Journal of Sports Medicine, 2004The 79% figure, the knee/back/shoulder breakdown, and the time-loss injury rate.
Primary A pilot biomechanical assessment of curling deliveries: is toe sliding more likely to cause knee injury than flatfoot sliding? Sports Biomechanics, 2017The moment-arm comparison between the two styles.
Study of curling mechanism by precision kinematic measurements J. Murata, 2022For the knuckleball behaviour of a stone with almost no rotation, and the recommendation to use a deliberate slight rotation.
Primary The Rules of Curling, R5: Delivery World Curling, 2025The hack rule, the release-before-the-hog-line rule and what happens when either is broken.

What a first session is actually like Learn the weight scale