About Fastest Slide
A bead slides without friction down a wire from a start gate to a finish post lower down. Which shape of wire gets it there first? The straight line is the shortest way, but the bead spends a long time on it going slowly. A curve that drops steeply first builds up speed early and wins, even though it is longer. The fastest of all is a cycloid, the curve traced by a point on the rim of a rolling wheel, and when the finish is far enough across it dips below the finish and climbs back up to it.
RACE: tick the tracks you want and press RACE. Drag the finish post to move it; press DRAW YOUR OWN and drag a curve from the start to try one of yours (if part of it rises above the start, the bead can never get past it). The speed buttons slow the race down to watch it; the stopwatches always show the true times for a board this size (the finish starts 1.2 m across and 35 cm down). ROLLING BALL swaps the bead for a solid ball rolling without slipping: every time goes up by the same factor, the square root of 7/5 (about 1.18), so the order of the finish never changes.
SAME TIME: four beads on a bowl shaped like an upside-down cycloid, each on its own wire, let go together from different heights. They all reach the bottom at the same moment, whatever the height. Switch to a CIRCLE bowl (an ordinary pendulum) and the ones let go from higher up arrive later. Drag a bead to change where it starts. HUYGENS' STRING hangs the first bead on a string from two curved cheeks, the way Huygens made a pendulum swing on a cycloid.
WHY: Johann Bernoulli's own argument. Light travelling between two points takes a path whose time is stationary (Fermat's principle); in layers like these that path is the quickest one, and it bends where its speed changes (Snell's law). Fill the board with layers of glass in which light goes as fast as a falling bead would at that depth, and the ray is the quickest path through those layers. With a few layers it is only a rough copy of the fastest slide; the more layers, the closer it comes, and with endlessly many thin layers it is exactly the cycloid. The readout gives the ray's widest gap from the cycloid: the largest distance, anywhere along the ray, to the nearest point of the cycloid. The second picture rolls the wheel that draws the cycloid for the race's finish post.
THE STORY
In Two New Sciences (1638) Galileo showed that a bead gets to the bottom of a circle sooner along the arc than along the straight chord, and went on to suggest the arc of a circle is the quickest path of all. The first part is right; the second is not.
Johann Bernoulli set the problem as a public challenge in the journal Acta Eruditorum in June 1696, and named the curve the brachistochrone, Greek for "shortest time". Leibniz persuaded him to give more time so that mathematicians abroad could try. Five solutions came: from Johann himself, his brother Jakob, Leibniz, the Marquis de l'Hôpital and Isaac Newton. Newton, by the account of his biographer John Conduitt, solved it in one evening; his answer was printed without his name in the Royal Society's Philosophical Transactions in January 1697, and Bernoulli is said to have recognised the author "as the lion by its claw". Johann's, Jakob's and Leibniz's solutions, with a translation of Newton's, appeared in Acta Eruditorum in May 1697; l'Hôpital's was not printed until 1988. The problem helped start the branch of mathematics now called the calculus of variations.
Christiaan Huygens had already found in 1659 that the cycloid is also the tautochrone, the "same time" curve, and published the proof in Horologium Oscillatorium (1673). An ordinary pendulum keeps slightly different time for big and small swings; Huygens hung his pendulum between curved cheeks so that the bob swung on a cycloid. In practice the cheeks did not help much: the bending string added friction, and other errors in the clocks were larger.
HOW IT WORKS
Each track is a chain of short straight pieces (hundreds for the curves). On a straight piece the bead's acceleration along the wire is constant, g times the drop divided by the length, so the time on each piece comes from solving a quadratic: no small time steps, and the start, where the bead has no speed yet, is handled exactly. At every corner the speed is set from the height fallen, v = √(2gh), so no energy is lost or gained. A bead can slow, stop, turn back and swing, which is what happens on a curve that rises above the start. On the cycloid this gives the textbook time, θ√(r/g) for a wheel of radius r turned through θ, to within a few parts in a million.
The cycloid through the start and the finish is found by solving (θ − sin θ) / (1 − cos θ) = across / down for the angle θ the wheel turns. When across / down is more than π/2, θ is past half a turn and the curve goes below the finish. Galileo's arc is the circle whose lowest point is the finish. The steep dive falls straight down, turns a tight bend and runs level to the finish.
On the cycloid bowl the pull along the wire is proportional to the distance left to the bottom, so the bead moves exactly like a spring: the time to the bottom, π√(r/g), doesn't depend on where it starts. The circle bowl has radius 4r, so tiny swings take the same time as on the cycloid. Bigger swings take longer; their times come from the same piece-by-piece run, which was checked while building the page against the pendulum's exact formula (an elliptic integral) and agrees to a few parts in a hundred thousand.
The light picture uses flat layers of equal thickness, each with the speed of a bead at its middle depth. The page picks the starting angle so that the ray arrives at the finish.
Simplifications: no friction or air drag, a bead the size of a point that never leaves the wire, and g = 9.81 m/s². A real ball rolls on two rails, not a flat surface, which slows it by a bit more than the factor above; real ball-on-track races also lose a little to friction, which can change the order when the times are close. With friction the fastest curve is no longer exactly a cycloid.
FURTHER WATCHING AND READING