Reading the Force Curve: What Repeatability Actually Means

A force curve is a graph of handle force against time for a single stroke, and that is the whole of what it is. The temptation is to read it as a verdict: this peak is too late, that plateau means a hang, this rounded shoulder is a rounded back. Some of that is real. A lot of it is not, because the curve is one sample from a movement that varies, and because the number the monitor reports depends heavily on which part of the stroke you ask it to measure.

In one study, the same rowers on the same unmodified ergometer produced 423 watts when asked for a full slide-drive cycle and 370 watts when asked for the drive alone ([4] Held 2022). That fourteen percent is not a difference between rowers. It is a difference between two questions asked of one person.

What makes a force curve worth reading is not its shape. It is whether the shape comes back.

The context

The PM5 resolves handle force across the stroke and derives the per-stroke fields most rowers end up looking at: peak force, drive time, drive length, and handle speed ([2] Concept2 2024). Those fields come with any machine carrying the standard monitor, which is a large part of why the curve has become the default technique feedback. But they are not independent measurements. Drive length and handle speed are the distance and the velocity of the handle during the drive, and drive time is how long that takes, so a curve that moves one of them has usually moved the others.

The stroke has a defined sequence — catch, drive, finish, recovery — and the force trace is concentrated in the drive ([3] Concept2 2024). That concentration is not a design choice; it follows from the mechanics. Rowing is a high-force, high-velocity whole-body problem, which is exactly the problem a force-velocity curve is built to describe.

Nothing about a single stroke is a verdict on a rower. A curve is one draw at one moment, and every stroke in a set differs from the last because the body does not reset between pulls. What a single curve can do is give you a trace of what happened during that draw, in enough detail to reason about. What it cannot do is tell you whether that draw is typical. Only a set of strokes can do that, and only once the rate and the stroke type are held still.

Coaching material has names for the shapes you will see: a sharp early peak, a rounded or plateaued top, a peak arriving late in the drive, a curve that falls away before the handle is square. The temptation is to read each shape as a named fault. The evidence supports something narrower. What is well replicated is that rate changes performance, technique and core stability on the ergometer together ([5] Duchene 2025), and that rate and intensity act on rowing technique as a pair ([6] Ettema 2025). A shape that looks like a fault at 18 spm may be the expected shape at 32, so "this rower has a late peak" is very often "this rower was not on the rate you compared against".

So a force curve is a measurement, taken during a movement, of a machine. Each of those three words carries a limit, and the limits are where most of the disagreement about curves comes from.

The core idea

The most useful thing about a rowing force curve is that it is not merely a shape. It is a force-velocity measurement, sampled stroke by stroke.

The reference study is eleven collegiate male rowers on an inertial load ergometer, asked to produce five maximal strokes at different loads. Pull force against average handle velocity was close to linear, with r² of 0.99, and instantaneous peak force against velocity gave r² of 0.93. Power against velocity, by contrast, was quadratic — r² of 0.98 to 0.99 — with an apex rather than an endless rise, at roughly 2.04, 3.25 and 3.43 metres per second depending on which power measure you take ([1] Sprague 2007).

Two things follow, and both are easy to get wrong in conversation.

The first is that rowing power-velocity has a maximum. There is an optimal handle speed, and going faster past it costs power rather than gaining it. That is why the rate you choose changes what the curve can show you, and why a curve recorded at one rate has no standing at another.

The second is subtler. Those rowers were free to choose their own duty cycle — the share of the stroke spent pulling — and duty cycle influences average power. So a force curve measures a rower and a stroke shape, and the stroke shape is partly a free parameter. A curve measured on a slow, long draw is not comparable to one measured on a fast, short one, even for the same athlete.

There is a practical consequence hiding in the apex. If power against velocity turns over, then going faster is not a free way to get more power — past the optimum it is a way to give some up. The rate that maximises your power is a property of you and of the stroke you are willing to row, not a free choice you make once. That is also why a curve recorded at a rate you would never race at is a poor thing to argue about.

That is the practical version of a methodological point: a force curve is only interpretable at a known rate. A single-stroke reading taken at whatever rate happened to be going is not a neutral observation, it is an observation of a movement you did not intend.

The how-to

A procedure that survives the evidence above.

Hold the rate constant. Record the rate with the curve. If you are comparing two sets of strokes, they have to be at the same rate, or you are comparing two movements.

Hold the stroke type constant too. Decide whether you are measuring a drive or a full slide-drive cycle, and never compare one against the other. The same rower is 14% apart between them ([4] Held 2022).

Judge a set, not a stroke. Ten or more strokes, then look at the spread and treat the coefficient of variation as the result. In the fifty-two-rower study, intraclass correlations ran from 0.94 to 1.00, standard error of measurement stayed at or below 2.7%, coefficients of variation at or below 4.9%, and limits of agreement within about 30 watts — for both stroke types and across a week ([4] Held 2022). Under about five per cent, you are inside the reliability of the instrument.

Change one thing and watch the spread, not the shape. Foot-stretcher position changes ergometer rowing kinematics, and it does so in a way a rower may not notice happening ([7] Engstrom 2023). A curve that is stable until you move the foot stretcher was never a property of your technique.

Ask whether the number predicts anything. Repeatability is not the goal. A measurement that repeats perfectly and means nothing is a tidy way to be wrong. A race-specific peak-power protocol has been shown to track both 2,000 m and 1,500 m elite performance, with between-session intraclass correlations of 0.92 to 0.99 ([11] Astridge 2025) — which is the test a repeatability figure on its own cannot pass.

The trade-offs

Three honest limits.

A repeatable curve can still be the wrong target. A curve can be beautifully repeatable — rising at the same angle, peaking at the same point, every single rep — and be the wrong target, because the rower is consistently doing the wrong thing. External focus of attention produces more fluent and more regular movement execution than internal focus, with the regularity itself being the measurable outcome ([8] Kal 2013). Regularity is not quality. It is evidence that a habit has formed, and habits hold on to bad technique just as well as to good.

The ergometer is not the boat. A handle-force curve from an ergometer measures a different movement from the one you race. That is not a criticism of the ergometer; it is a statement about what is measured. Three on-water instrumentation systems were compared directly against a Concept2 for concurrent validity precisely because the agreement could not be assumed, and the ergometer's stroke-to-stroke variation in power was far wider than the reference system's ([9] Holt 2021). On-water force-velocity relationships have since been characterised directly: maximum force on the ergometer ran 23% higher and maximum velocity 31% lower than on the water, and the two velocities were not significantly correlated at all ([10] Nordez 2025). An ergometer curve is evidence about the ergometer.

Some variation is healthy. Physiological complexity in muscle force and torque changes with ageing, disease and neuromuscular fatigue, and a body that has lost complexity has lost adaptability ([12] Pethick 2021). A little stroke-to-stroke variation is a healthy nervous system, not a fault. The question is never "is this variable" but "is this variable compared with what".

What to do tomorrow

Pick one piece, settle on a rate, and row ten strokes recording the same fields each time. Then do it again a week later.

Record what you changed and when. A curve with a note attached — feet at 7, 26 spm, second piece — is worth something in three months. The same curve with no note is a number you will re-derive from scratch and get slightly differently each time, because the two measurements were never comparable in the first place.

Look at two things: the spread across your ten strokes, and whether the rate you rowed at is the rate you meant to row at. Most people who take their force curve seriously find one of those two is wrong, and it is usually the second.

If the curve moves when you change your foot stretcher, you were reading the machine ([7] Engstrom 2023). If the spread is wider than you expected, check the rate before you check your technique ([5] Duchene 2025). If both look right and the stroke still feels wrong, the force curve was never going to be the thing that told you.