Every muscle has one length where it is strongest, and two different ways to be weak

The pull comes from overlap

A muscle does not squeeze. It ratchets.

Inside every muscle fibre, two sets of protein filaments lie side by side. The thick ones carry myosin, a motor protein whose heads reach out. The thin ones are built from actin. Each myosin head grabs the thin filament, tilts, lets go, and grabs again [1]. The tilt is tiny: roughly ten nanometres of sliding, at about five piconewtons of force, per head [1]. One stroke moves nothing you could feel.

Force comes from doing it in parallel. A muscle fibre is built from millions of copies of one repeating unit, called a sarcomere. Myosin motors stick out in two arrays from the centre of each thick filament, and thin filaments run in from both ends of the sarcomere [1]. Cryo-electron microscopy and x-ray diffraction have now mapped how those heads sit along the filament surface [3][6]. The model that emerged over sixty years is a tilting lever, repeated by the million [2].

So the force a muscle can make right now depends on one thing above all. How many heads currently have something to grab. That is set by how far the two filament sets overlap [4]. Overlap is a function of length.

Two ways to be weak, for opposite reasons

Stretch a muscle far enough and the filaments slide apart. Fewer heads reach the thin filament, so fewer can pull [4][5]. Physiologists call this the descending limb of the force-length curve.

Shorten it far enough and the opposite happens. The thin filaments coming in from each end run into each other. Further shortening now reduces force, because too much overlap makes the thin filaments interfere and distorts the internal structure [4]. That is the ascending limb.

Between them sits a plateau, the length where overlap is greatest and the muscle makes its peak force [4]. A review published this year states the same shape plainly. Contractions at short muscle lengths, which usually mean a bent joint, involve poorer overlap and fewer working cross-bridges, the temporary grips between the two filaments. Contractions at moderate lengths get better overlap and more force [5].

At long lengths a third thing joins in. A protein called titin runs the length of the sarcomere and behaves like a spring [7][8]. Titin is the largest protein in the human body, spanning the sarcomere from one anchoring disk to its centre line [7]. Under stretch its coiled sections unfold and it resists [8]. Pull a mouse muscle from about 2.4 to 2.7 micrometres of sarcomere length and passive force rises as titin stretches [9]. A very long muscle is weak in its active grip and stiff in its passive one.

Your skeleton has its own opinion

Now put that muscle on a bone.

What you can lift is not muscle force. It is torque: force multiplied by the perpendicular distance from the joint’s centre to the line the muscle pulls along [12][13]. That distance is the moment arm, and it changes as the joint rotates [13]. So a joint angle changes two things at once. How much overlap the muscle has, and how much leverage it gets.

The combined effect is measurable and large. Sixteen adults performed hip adductions while the muscle was being lengthened under load. Peak torque was significantly lower at 90 degrees of hip flexion than at 0 or 45 degrees [10]. Electrical activity in the muscle, measured at the skin, did not differ across the three positions [10]. They were trying just as hard. The position gave back less.

Geometry alone can do it. Thirteen lifters were filmed deadlifting at 60 to 90 percent of their maximum. Rounding the lower back shortens the back extensors’ moment arm, which reduces the torque they can produce [14].

In a gym this shows up as the sticking point, the weakest part of a lift. It is defined precisely as the stretch of the bar’s path between its first peak speed and the next dip [15][16]. A fixed weight does not care about your curve. Cams, bands and chains have been used to raise resistance across that stretch instead. Roughly 98 percent of the constant load before the sticking region, 105 percent inside it, 113 percent after [15].

The strongest length is not fixed

The tidy version says each muscle has one optimal length. Real muscles are less tidy.

Researchers measured the jaw muscle of six rats biting food of different hardness, in life, and mapped it onto that muscle’s own force-length curve [4]. It worked from 7 percent shorter than its optimum out to 27 percent longer. That is well onto the descending limb, where the consensus said muscles avoid working [4]. More striking, the optimum itself moved. At submaximal effort it sat about 12 percent shorter than the optimum measured from a single twitch [4].

Human muscles disagree with each other too. Fifteen men held steady knee bends at three angles. Maximum force was highest at the long muscle length, but control was worst in the middle, where their torque wobbled most [11].

The curve also shifts with training. Work done while a muscle lengthens under load is associated with longer resting muscle bundles, and with the angle of peak torque moving toward longer lengths [10][20]. The change is slow and still argued over. One line of work found ten days of downhill running were needed before the number of sarcomeres in series rose. Five days did not do it, and a single session did not either [21].

Where the measuring stops

Almost none of this was measured on one muscle in one living person.

A joint moment is the sum of every muscle crossing that joint. Each has its own architecture, geometry and moment arm, and each works against its own opposing muscles [17]. Measuring one muscle’s force directly means putting a sensor on its tendon during surgery [17]. Everything else is estimated from models, and those models assume things, including where each muscle’s optimum length sits [17].

The assumptions matter. Personalising joint positions and muscle paths from MRI scans substantially changed the calculated moment arms in children with a rotated thigh bone [19]. Moment arms help set how much torque a muscle can make [19]. A recent methods paper makes a narrower point about the lab bench itself. Even the starting length chosen for a test leaves the filaments at inconsistent overlap, which changes how fast force rises [18]. There is no single correct protocol [18].

Which is why the practical questions stay open. An eight-week trial compared full-range with partial-range elbow training held at long muscle lengths. The long-length group grew slightly more in the lower biceps, but the effect sizes were trivial to small, and full range improved one-rep maximum more [22]. Where in its range a muscle works appears to matter. How much is still being argued.