Mechanics

The walk that has to become a run

In walking the body vaults over a stiff leg, and the hip travels on an arc centred on the foot. Moving on an arc takes an acceleration towards its centre, and a foot can push on the ground but not pull on it, so the only thing that can supply that acceleration is gravity. That sets a speed no walker can exceed — the square root of g times the leg length, three metres a second for an adult, one point two on the Moon — and it is why people break into a run near two metres a second, why the Apollo astronauts hopped, and why a fossil trail of footprints records how fast the animal that left it was going.

Assumes: Turning is an acceleration, and constant speed does not help · The pendulum, and the small lie that makes it simple

Speed up a walk and at some point it stops being a walk. Nobody decides to run; the gait changes by itself, at a speed that is remarkably consistent for adults — a little over two metres a second — and children, with shorter legs, change at a lower speed and spend most of a walk with an adult half-running to keep up. Race walkers, whose rules forbid running, reach four metres a second by a gait so contorted it looks like a different animal. On the Moon, where everything weighs a sixth as much, astronauts who needed to cover ground did not walk at all; they hopped.

All of this comes from a constraint that has nothing to do with muscles or tiredness. A walking body travels, at each step, on an arc. An arc needs an acceleration towards its centre. And the one thing that can provide it, in walking, is gravity.

The hip on an arc

In a walking stride the leg that bears the weight is nearly straight, and the body vaults over it like a pole-vaulter over a stiff pole: the hip rises as the leg passes vertical and falls as it tips forward. With the foot fixed on the ground, the hip moves on a circle of radius LL, the leg’s length, centred on the foot. Turning is an acceleration established that a body moving on a circle at speed vv has an acceleration v2/Lv^2/L directed towards the centre, whether or not its speed is changing, and that something must supply the force.

Here the centre is the foot, below the hip, and the force towards it is the body’s weight minus whatever the ground pushes up. At the top of the arc, with the leg vertical, the balance is

N=m(g−v2L),N = m\left(g - \frac{v^2}{L}\right),

where NN is the force between foot and ground. The faster the hip vaults the foot, the less the ground has to hold up, because more of the weight is being spent on bending the hip’s path round.

The faster the hip vaults the foot, the less the ground has to hold up. The force between foot and ground at the middle of a stride, as a fraction of body weight, against walking speed, for a leg 0.9 m long on Earth, Mars and the Moon. Carrying the hip over the foot on an arc needs a centripetal acceleration v²/L, which only gravity can supply, so the ground force is m(g − v²/L). It reaches zero at √(gL): 2.97 m/s on Earth, 1.83 on Mars, 1.21 on the Moon. Past that the walker leaves the ground at the top of every step, which is to say runs.
Fig. 1 The force between foot and ground at the middle of a stride, as a fraction of body weight, against walking speed, for a leg 0.9 m long on Earth, Mars and the Moon. Carrying the hip over the foot on an arc needs a centripetal acceleration v2/Lv^2/L, which only gravity can supply, so the ground force is m(g−v2/L)m(g - v^2/L). It reaches zero at gL\sqrt{gL}: 2.97 m/s on Earth, 1.83 on Mars, 1.21 on the Moon.

At v=gLv = \sqrt{gL} the ground force reaches zero. Above it the equation asks for a negative force — the foot would have to pull the body down onto it — and a foot cannot pull. What happens instead is that the body leaves the ground at the top of the arc, and a gait with a moment of flight in each stride is a run. For a leg of ninety centimetres the limit is 2.97 metres a second, a little under eleven kilometres an hour. Nobody walks that fast for the reason the equation gives: no arrangement of muscles can make a stiff-legged vault hold the ground beyond it.

Seen from the walker’s own hip the same limit has a different description, and the two have to agree. In a frame that swings with the leg, the hip is at rest at the top of the arc, and to keep it at rest the bookkeeping adds an outward force mv2/Lmv^2/L — the centrifugal force that the forces that are not there introduced as the price of describing motion from a turning frame. The walker in that frame feels lighter at the top of each step, by exactly that amount, and at gL\sqrt{gL} feels weightless. Both descriptions predict the same lift-off at the same speed; they differ only in which side of the equation the v2/Lv^2/L is written on.

The same arithmetic is the reason the ball a turntable will not throw off and a car going over a humpbacked bridge share an experience with a walker. A car crossing the crest of a bridge of radius RR loses contact at gR\sqrt{gR}; the hump in a walker’s path is the leg itself, and the leg is a much smaller radius than any bridge, so the walker reaches the limit at a pedestrian speed. Roller-coaster designers meet the limit from the other direction: a hump taken too fast lifts the riders out of their seats, and the loop that is not a circle showed why the tracks over the top are shaped so that the curvature, and with it the demand on gravity, changes gradually rather than at once.

Gravity sets the pace, and the Moon is slow

The walking ceiling depends on two things, the leg’s length and the strength of gravity, and on nothing else: not on the walker’s mass, not on muscle strength, not on how the leg is built. That makes it easy to move to another world.

On the Moon the fastest walk is a stroll. The highest speed at which a 0.9 m leg can still walk, √(gL), and the speed at which people actually change to running, where v²/gL is about 0.5, against the strength of gravity. Both scale as the square root of g. On the Moon walking runs out at 1.21 m/s and the change to a run would come at 0.85 m/s, a slow amble — which is why the Apollo astronauts, who needed to cover ground, hopped and loped instead of walking. Experiments in which people walk with part of their weight carried by a harness find the change near the same ratio.
Fig. 2 The highest speed at which a 0.9 m leg can still walk, gL\sqrt{gL}, and the speed at which people actually change to running, where v2/gLv^2/gL is about 0.5, against the strength of gravity. Both scale as the square root of g. On the Moon walking runs out at 1.21 m/s and the change to a run would come at 0.85 m/s, a slow amble.

On the Moon, at a sixth of Earth’s gravity, the ceiling falls by the square root of six to 1.21 metres a second, an unhurried stroll. The weight is lighter, but the weight is exactly what was holding the walker to the arc, and with less of it the arc can be traversed only more slowly. The Apollo astronauts discovered this in practice: on the films they move by a loping, two-footed hop or a bounding run, because a walk on the Moon is so slow it is barely worth doing, and a run there costs little since every landing is soft. On Mars the ceiling is 1.83 metres a second, and future visitors will find a brisk walk turning into a jog much sooner than at home.

The observed change of gait comes well before the ceiling. People switch to running at a speed where v2/gLv^2/gL — the Froude number, the ratio the ceiling sets to one — is about a half: 2.1 metres a second on Earth, 0.85 on the Moon. Rodger Kram and colleagues tested this in 1997 by having people walk on a treadmill while a harness carried part of their weight, simulating lower gravity, and found the change of gait moving to lower speeds, as the square root of the effective gravity predicts. Why the change comes at a half rather than at one is not settled by this argument alone; the leading explanations are that walking near the ceiling becomes mechanically costly, as the legs have to work harder to keep the arc smooth, and that the muscles’ own force–speed limits begin to bind. But the scale of the speed is set by gL\sqrt{gL}, and so is its dependence on gravity.

A longer stride lowers the ceiling

The top of the arc is not the hardest point. Away from the vertical, at heel strike and at toe-off, the hip is lower and moving faster — it has fallen from the top, trading height for speed — so it needs more centripetal acceleration, while the component of gravity along the leg, which is what supplies it, is reduced by the cosine of the leg’s angle.

A longer stride lowers the ceiling. The highest speed at the top of the arc at which the foot still presses on the ground throughout the step, as a fraction of √(gL), against the angle the leg makes with the vertical when the foot lands. Away from the vertical the hip is lower and moving faster, so it needs more centripetal acceleration while gravity supplies less of it along the leg: the condition is that the top speed squared, over gL, stay below 3 cos θ − 2. At 20°, a normal stride, the ceiling is 0.91 of √(gL); at 48.2° it is zero. Speed in walking comes from step rate more than from step length for this reason.
Fig. 3 The highest speed at the top of the arc at which the foot still presses on the ground throughout the step, as a fraction of gL\sqrt{gL}, against the angle the leg makes with the vertical when the foot lands. The condition is that the top speed squared, over gL, stay below 3 cos θ − 2. At 20°, a normal stride, the ceiling is 0.91 of gL\sqrt{gL}; at 48.2° it is zero.

Putting both effects into the balance gives the condition that the top speed squared, divided by gLgL, stay below 3cos⁡θ−23\cos\theta - 2, where θ\theta is the leg’s angle at landing. For a normal stride, with the leg twenty degrees from vertical at heel strike, the ceiling is 0.91 of gL\sqrt{gL}; for a very long stride it falls further, and at 48.2 degrees there is no speed at which the foot stays loaded throughout. A walker going faster therefore gains less from lengthening the stride than intuition suggests, and real walking speeds rise mostly through faster stepping. The same formula, with the 3 and the 2, is the condition for a bead sliding off the top of a smooth sphere — it leaves at the height where cos⁡θ=2/3\cos\theta = 2/3 — and the walker’s leg is that sphere’s radius.

Two shapes of the same physics

The walker’s ceiling is a ratio of a speed squared to gravity times a length, and the same ratio governs a ship. The wave that holds a ship back found a displacement hull’s practical speed limited near gL\sqrt{gL} times a constant, with LL the hull’s length, because the ship makes a wave whose wavelength grows with speed until it is as long as the ship and the bow climbs its own bow wave. The number is the Froude number in both cases, named for William Froude, who tested model ships in the 1870s and found that a model and a full-size ship behave alike when this ratio matches. The walker’s leg and the ship’s hull are different physics — an arc that gravity must bend, a wave that gravity must make — but both are motions in which gravity is the only force that can supply what the motion needs, and both are limited at the speed where the motion asks more of it than it has.

That is what makes the ratio useful for animals. If gravity sets the pace of walking, animals of very different sizes should move alike when their Froude numbers match: a cat and a horse, each at a Froude number of a quarter, should be at the same point in their gait, with geometrically similar strides. R. McNeill Alexander tested this across mammals from rodents to rhinoceroses in the 1970s and found that it held well — dynamic similarity, the walking version of Froude’s ship models. It is a different scaling from the one the jump that does not get higher with size found for leaping, where muscle energy grows with mass and so does the weight to be lifted, leaving every animal’s jump about the same height. Walking speed does depend on size, as the square root of the leg’s length: an elephant’s comfortable walk is faster than a mouse’s, but not by the factor of their lengths.

A footprint trail records the Froude number, and so the speed. Stride length divided by hip height against the Froude number v²/gh, in R. McNeill Alexander's empirical fit to walking and running animals from people to elephants: λ/h = 2.3 times the 0.3 power of v²/gh. Animals of every size moving at the same Froude number take geometrically similar strides, which is what dynamic similarity predicts if gravity sets the pace. Turned round, the fit reads a speed from a fossil trackway: a stride of 2 m left by an animal whose hip stood 1 m high implies about 2.5 m/s. The walking range ends near a relative stride of 1.9.
Fig. 4 Stride length divided by hip height against the Froude number v2/ghv^2/gh, in R. McNeill Alexander’s empirical fit to walking and running animals from people to elephants: λ/h = 2.3 times the 0.3 power of v2/ghv^2/gh. Turned round, the fit reads a speed from a fossil trackway: a stride of 2 m left by an animal whose hip stood 1 m high implies about 2.5 m/s. The walking range ends near a relative stride of 1.9.

Alexander’s fit has a use he intended from the start: reading speeds off fossil trackways. A trail of footprints records the stride length directly, and the size of each print gives an estimate of the hip height, roughly four times the foot length for many dinosaurs. The ratio of the two gives the Froude number, and the Froude number with the hip height gives the speed. A biped whose stride was twice its hip height was moving at about two and a half metres a second for a hip a metre high. Applied to real trackways, the method shows that most dinosaur trails were left by animals walking at a few kilometres an hour, as most modern animal tracks are, with a handful recording runs. The fit is empirical and the hip-height estimate rough, so a trackway speed is good to perhaps thirty per cent, but no other method gives any speed at all for an animal dead a hundred million years.

Why walking is cheap until it is not

The vaulting arc is not only a constraint; it is the reason walking costs so little. As the body rises over the stance leg it slows, turning kinetic energy into height, and as it falls forward beyond the foot it speeds up again, turning the height back into motion.

Walking trades speed for height and back, twice a stride. The kinetic energy and the gravitational energy of a 70 kg walker's centre of mass over one stance, with the leg sweeping from 20° behind to 20° in front and 1.25 m/s at the top, for a rigid 0.9 m leg. The body rises 37 J worth and slows to 55 J of kinetic energy at mid-stance, then falls forward and speeds up; the two trade almost exactly, as in a pendulum turned upside down. The muscles need only make good the loss at each heel strike, which is why walking is cheap — and why it fails when the arc can no longer be held to the ground.
Fig. 5 The kinetic and gravitational energy of a 70 kg walker’s centre of mass over one stance, with the leg sweeping from 20° behind to 20° in front and 1.25 m/s at the top, for a rigid 0.9 m leg. The body rises 37 J worth and slows to 55 J of kinetic energy at mid-stance, then falls forward and speeds up; the two trade almost exactly, as in a pendulum turned upside down.

For a seventy-kilogram walker at a comfortable 1.25 metres a second the body rises about 37 joules’ worth over each stance and trades it back, in step with the kinetic energy, so that the total barely changes. It is the pendulum and its small lie turned upside down — an inverted pendulum, carrying energy between two forms with no one having to supply it — and it is the same exchange the hill that gives it back followed for a ball rolling over a bump. Nothing outside the walker supplies energy on level ground — the floor that does no work found the ground pushing on a foot that does not move while it pushes — so the exchange is internal bookkeeping, and the muscles have to make good only what is lost at each heel strike, when the body’s velocity is redirected from falling to rising again, and the measured exchange in human walking recovers about two-thirds of the energy that would otherwise have to be supplied. The swinging leg adds a second pendulum: left to itself it swings forward in about the time a step takes, so a comfortable walking pace is close to the leg’s own natural period. A swinging leg is a compound pendulum whose period grows as the square root of its length, the same dependence that made the length nobody has to measure a way of finding a pendulum’s length from its beat, and the natural stepping rate of a long-legged animal is slower in exactly that proportion. The step rate falls as g/L\sqrt{g/L} and the step length grows as LL, so the natural walking speed grows as gL\sqrt{gL} — the same combination as the ceiling, arrived at by a different route.

The exchange works only while the hip stays on its arc. Near the ceiling the arc’s demands on gravity grow, the body has to be pulled down onto the leg at the top, and the muscles start paying for what the pendulum had been doing free; in running the body switches to a different exchange altogether, bouncing on legs that act as springs, storing the energy of each landing in stretched tendons and returning it at take-off. Walking is an inverted pendulum and running is a pogo stick, and the walk-to-run transition is the speed at which the cheaper of the two changes.

Race walkers, children and the edges of the argument

Race walkers are the test of the ceiling, because their rules require the walker to stay in contact with the ground. Elite men cover twenty kilometres at about 4.2 metres a second, well above the three metres a second a straight-legged vault allows. They do it by not vaulting: the hips rotate and drop so that the body’s centre of mass travels on a much flatter path than the hip joint, with an effective radius of curvature closer to two metres than to one. The rules’ requirement of a straightened knee is an attempt to keep the gait a walk, and judges watch for loss of contact by eye; high-speed film shows elite race walkers in the air for a few hundredths of a second per stride, under the eye’s resolution. The ceiling is not broken, it is moved by changing the arc.

Children are the other edge. A child with a fifty-centimetre leg has a walking ceiling of 2.2 metres a second and changes gait near 1.6, so a parent striding along at 1.7 is above the child’s preferred change, and the child trots. The Froude number explains why children seem to run everywhere: at an adult’s walking pace, they are running by the same physics.

Everything here treats the leg as a rigid strut and the body as a point at the hip. Real legs flex at the knee during stance, real hips are not the centre of mass, and the foot rolls from heel to toe rather than pivoting on a point, all of which change the arc’s effective radius by a few tens of per cent. The transition at a Froude number of a half is an empirical fact that the arc argument frames but does not derive. And the stride fit is a curve through data from many species, accurate to a few tens of per cent for any one animal.

Still open: why the change comes where it does

The ceiling is mechanics and is settled; the transition below it is not. Proposals for why people switch at a Froude number of a half rather than nearer one include the metabolic cost of each gait crossing over, the forces on particular muscles or the ankle’s flexors reaching a limit, and the stability of the gait; experiments that load or unload different parts of the body, or change the cost of one gait while keeping the other, have found support for more than one, and the answer may be that several limits arrive near the same speed. Whether the same trigger works in four-legged animals, whose walk-to-trot transitions also cluster near a Froude number of a half, is an open comparison.

Walking carries the hip over the foot on an arc of the leg’s length, so it needs a centripetal acceleration v2/Lv^2/L that only gravity can supply, because a foot pushes and cannot pull: the ground force at mid-stance is m(g−v2/L)m(g - v^2/L), and it vanishes at gL\sqrt{gL} — 2.97 m/s for a 0.9 m leg on Earth, 1.21 m/s on the Moon — and lower still for a long stride, 0.91 of that at 20°. People change to running near half that ratio, 2.1 m/s; the same ratio limits a ship’s hull, matches animals of every size so that a fossil trackway can be read for speed, and is the price of the free energy exchange that makes walking cheap.

Part 10 of 10

This essay is one argument about Circular motion. The others:

The objects named here

The third axis, after the field and the reading path: the things themselves, and every essay that touches each one.

Centripetal accelerationDynamic similarityEnergy exchangeFroude numberInverted pendulum