How Does a Bicycle Stay Upright? Gyroscopes, Trail and a Century of Wrong Answers

A stationary bicycle falls over, and a moving one stays up for one reason: when it leans, it steers into the lean, and steering into the lean brings the wheels back under the centre of mass. A rider does this without knowing it, turning the bars a fraction towards whichever side the bike tips, so that the bike curves that way and the outward lean of the turn cancels the fall. The physics is the same as balancing a broom on a hand: the base is moved under the load. What makes a bicycle remarkable is that it does the steering by itself, without a rider, provided it is moving faster than a few kilometres an hour, and the question is what mechanism turns the front wheel into a lean.

A spinning wheel resists having its axis tilted and responds to a tilt by turning at right angles to it, which is precession, and a leaning bicycle's front wheel does precess in the direction that steers into the lean. The effect is real and it helps, and from a 1910 analysis onward it was taught as the reason. It is not sufficient. In 1970 the chemist David Jones built a bicycle with a second front wheel, mounted beside the first and spun backwards to cancel the gyroscopic effect exactly, expecting it to be unrideable, and found it rode almost normally and still balanced when pushed riderless, though a little less well. The gyroscope contributes; something else is doing most of the work.

The something else is the geometry of the front fork. On every bicycle the steering axis is tilted back, and if it is extended down to the ground it hits the road a few centimetres in front of the point where the tyre touches. That distance is the trail, and it means the front wheel is dragged behind its steering axis like a castor on a shopping trolley. Two things follow. When the bike leans, the weight on the front wheel, acting behind the axis, turns the wheel into the lean, which is the steering the balance needs. And when the wheel is turned, the front of the bike drops slightly, so that gravity itself pulls the steering towards the lean. Trail, with the gyroscopic effect assisting, was the accepted explanation from the 1970s, and the story would have ended there had anyone checked the equations.

In 2011 a team in Delft and Cornell published the results of a decade of working through the full equations of motion of a bicycle, which have 25 parameters and had been derived incorrectly several times since 1899. The equations said that neither gyroscopic effect nor positive trail was necessary, and to prove it the team built a two-mass-skate bicycle: small wheels spun by counter-rotating twins to cancel any gyroscope, a steering axis tilted forward so that the trail was negative, and weights placed so that the front assembly's centre of mass sat ahead of the rear's. Pushed across a gym floor at a jog, it balanced itself and recovered from a sideways shove. The general principle, which the equations reveal and no single mechanism captures, is that a bicycle is stable if a lean causes the front to steer into it faster than the rear falls, and there are several ways to arrange that:

• •Gyroscopic precession of the front wheel, which steers into a lean and matters more at speed
• •Positive trail, which turns a castoring front wheel into a lean under the bike's weight
• •Mass distribution, with the front assembly's centre of mass forward and low so that it falls into the lean first
• •Speed: below a critical speed none of these acts fast enough and the bike falls, above a higher speed it becomes too stable to turn easily, and in between it is self-correcting

A rider makes the machine stable at speeds where it would not be alone, by steering, and does so with an instinct so fast and unconscious that most riders describe the opposite of what they do. To turn left on a bicycle at speed, the rider first steers briefly right, which leans the bike left, and then steers into the lean; the move is called countersteering, motorcyclists are taught it explicitly, and cyclists do it without believing they do. The learning that takes a child a week is the learning of that reflex, and it is stored in a part of the brain that does not forget, which is why it is the thing one never forgets how to do. The mechanism that keeps the bike up, meanwhile, was designed by trial and error in the 1880s by makers who could not have said why their forks were raked, and their geometry is still what every bicycle uses.

Riding without hands shows the self-steering at work. With nobody holding the bars, the only thing a rider can do is shift their body, and moving the hips or leaning the shoulders tilts the frame slightly. The fork geometry then turns the front wheel into that tilt, exactly as it does on a riderless bike, and the bicycle curves that way. A rider who rides no-hands with ease is really letting the machine steer itself and nudging it with their weight, which is why the trick becomes noticeably harder with a heavy bag hanging from the handlebars, changing how the front assembly responds.

Slow riding shows the opposite. Below walking pace none of the self-stabilising effects acts quickly enough, so the rider has to make all the corrections, and the bike weaves in visibly larger swings as the front wheel is turned further to bring the wheels back under the body. That is why a child learning to ride wobbles most when going slowly, and why a push to gather speed makes everything easier. Track cyclists who balance almost motionless before a sprint are doing something different again, turning the front wheel sideways and rocking gently forwards and backwards so that the tyre's contact point shifts under them.

The first two-wheeler, built by Karl Drais in Germany in 1817, had no pedals at all: the rider sat astride it and pushed along the ground with their feet, gliding between pushes. Modern balance bikes for small children use the same idea, and they fit what the physics says about learning. Stabilisers, the small side wheels bolted to a child's first bicycle, hold the bike upright so it cannot lean, which means the child learns to steer it like a tricycle, turning the bars only to change direction and never linking a lean with a correction. When the stabilisers come off, that habit has to be unlearned.

A balance bike lets the child lean, wobble and steer back underneath themselves at walking pace, with feet ready to catch a fall, so the steering reflex develops first and pedalling is added later. How deeply the reflex is wired becomes obvious when it is reversed. The engineer and video maker Destin Sandlin built a bicycle whose handlebars turned the front wheel the opposite way, and found he could not ride it at all. Practising a little every day, it took him around eight months to master, and afterwards he briefly struggled to ride a normal bicycle. His young son, set the same task, learned much faster.

Early bicycles were hard and dangerous to ride. The high-wheeler, or penny-farthing, popular in the 1870s and 1880s, placed the rider high above a huge front wheel driven directly by the pedals, and a sudden stop could throw the rider head first over the handlebars. Riding one was a sport for young, daring men.

In 1885 John Kemp Starley in Coventry produced the Rover, often described as the first successful safety bicycle, with two wheels of similar size, a chain driving the rear wheel and the rider seated between them. Combined with pneumatic tyres, introduced soon after, it made cycling comfortable and safe enough for almost anyone. The basic layout of nearly every bicycle since comes from that design.

A moving bicycle stays upright because a lean makes the front wheel steer into it, bringing the wheels back under the rider, and it does that by itself through the gyroscopic precession of the front wheel, the castoring effect of trail in the fork geometry, and the forward placement of the front assembly's mass, none of which is essential alone, as a 2011 bicycle with no gyroscope and negative trail proved by balancing anyway. A rider adds steering, including the countersteer that starts every turn.