Pi is the ratio of any circle’s circumference to its diameter, about 3.14159, but the better answer to “what is pi?” is that π measures turning. Measure angles in radians, the natural unit in which an angle equals the arc it sweeps on a circle of radius 1, and a half turn is exactly π radians; a full turn is 2π. That is why π shows up in sine waves, pendulums, springs and even the bell curve, where no circle is drawn: anything that repeats can be given a phase on a circle, and π is the exchange rate between cycles and radians.
You were told that π is 3.14. That is an approximation, not an explanation.
Every circle, one number
Draw a circle. Measure its circumference, the distance around, and its diameter, the distance across through the center. Divide one by the other. The real claim hidden in the schoolbook definition is that every circle gives the same answer. Enlarge a circle ten times and both lengths grow tenfold, so their ratio stays put.
That shared number deserves a name, but naming it doesn’t calculate it. Wrap string around a jar and you might get 3.1, or 3.14 if you are careful. That is an estimate with uncertainty, not an exact value. Why near 3.14, rather than 3.2? To calculate digits, you need a procedure that traps the answer.
Archimedes squeezes the circle
Around 250 BC, in Syracuse, Archimedes wrote a short text called Measurement of a Circle. It has three propositions, and the third is a landmark: bounds on π, with a proof that the bounds hold.
The ratio of the circumference of any circle to its diameter is greater than 3 10/71 but less than 3 1/7. — Archimedes, Measurement of a Circle, Proposition 3, c. 250 BC
His strategy: curved lengths are difficult, straight lengths are manageable. Put one regular polygon inside the circle and another outside. The inner perimeter is too short, the outer perimeter is too long, and the circumference cannot escape.
Start with six sides. Each side of the inscribed regular hexagon equals the circle’s radius, so its six sides make three diameters. Its perimeter is 3 times the diameter, and π must exceed 3: the leading 3, secured by geometry rather than a ruler. The outer hexagon’s perimeter is 4√3 times the radius, about 3.464 diameters. That bracket is crude but certain.
Now double the sides: 12, 24, 48, 96. Each doubling uses square roots, and Archimedes bounded those roots by hand with fractions, keeping every approximation on the safe side. For √3 alone he used 265/153 < √3 < 1351/780, and he never explained how he found those fractions. At 96 sides, he stopped.
So π rounds to 3.14 at two decimal places, guaranteed. Archimedes knew that 22/7 is not π and made no claim to an exact value. π isn’t any polygon’s perimeter divided by its diameter. It is their common limit as the sides keep doubling: not a finished measurement, but an endless squeeze.
What pi really measures: half a turn
Set a circle’s radius to 1 and move a point around its edge, watching the radius turn. The arc length traveled equals the angle swept out, measured in radians. That is how the radian is defined: one radian is the angle whose arc is as long as the radius. A quarter turn is π/2. A half turn is π. A complete turn is 2π.
So π isn’t merely a decimal attached to a circumference. It measures the turn that points you in the opposite direction. The radian is the unit; π radians is the reversal. The word is younger than the idea: “radian” first appeared in print on 5 June 1873, in examination questions set by James Thomson at Queen’s College, Belfast.
Euler’s formula makes the reversal compact. On the complex plane, multiplying by eiθ rotates through θ radians. Turn by π, and 1 lands on −1: eiπ = −1. Why the number e belongs there is its own story, told in Why e is 2.718….
Now track the rotating point’s height while moving the drawing sideways, which is how the video unrolls it: a sine wave appears. It repeats every 2π in its angle and crosses zero at every multiple of π. Rotation has become a wave.
A pendulum is another disguise. For small swings, its period is approximately 2π√(L/g), where L is the length and g the gravitational acceleration. The bob retraces an arc, not a complete circle, but its phase advances through one full cycle each period. That cycle is 2π radians, the same phase bookkeeping used for springs, alternating current, planetary orbits and vibrating molecules. A repeating motion can be assigned a phase on a circle even when nothing physically goes around.
Even the bell-shaped curve e−x² has total area √π. Square that integral and switch to polar coordinates, and a hidden circular symmetry explains why π appears. Roundness was there all along, in disguise.
What kind of number is π?
Counting numbers sit inside the integers, integers inside the rational numbers, and rational numbers inside the real numbers. Each larger set admits numbers the smaller one leaves out. Where does π belong? Archimedes’ upper bound, 22/7, is rational. So is 355/113, which shares π’s first six decimal digits. For roughly two thousand years the question remained open: could π itself be a fraction nobody had found?
In 1761 the Swiss mathematician Johann Heinrich Lambert, of Mulhouse, supplied the answer: no. (The memoir appeared in the Berlin Academy’s volume dated 1761, printed in 1768, which is why some histories give the later year.) Using an infinite continued fraction for the tangent, he proved that the tangent of any nonzero rational angle, in radians, is irrational. But the tangent of π/4 equals 1, which is rational. So π/4 cannot be rational, and neither can π. Its decimal expansion never ends and never settles into a repeating block. That rules out repetition, not every possible pattern: irrationality does not prove that the digits are random.
Irrational numbers can still obey simple equations. √2 solves x² − 2 = 0, as the story of why the square root of 2 is irrational shows. Numbers that solve nonzero polynomials with integer coefficients are called algebraic; they include every fraction, but not every real number. In 1882 Ferdinand von Lindemann proved that π lies outside that family. No nonzero polynomial with integer coefficients has π as a root: π is transcendental. His proof built on Charles Hermite’s 1873 proof that e is transcendental, together with the fact that eiπ = −1. The half turn was part of the proof.
An ancient construction problem fell with it: squaring the circle with only an unmarked straightedge and a compass. A unit circle has area π, so an equal square needs side √π, and that length is transcendental too. Those tools construct only algebraic lengths from a unit segment, so the exact square is impossible. Archimedes hadn’t missed an algebraic shortcut. Exact formulas for π exist, but no finite combination of rational arithmetic and roots produces it. His endless squeeze remains one route to its digits.
A fact about flat space
One last twist. Take a sphere of radius R and center a circle at the North Pole, measuring its radius along the surface. Extend that surface radius a quarter of the way around the sphere, and your circle is the equator. Its circumference is 2πR. Its surface radius is πR/2. Divide the circumference by twice the surface radius and the ratio is exactly 2, not 3.14.
On a negatively curved, saddle-shaped surface, the opposite happens: circles can have circumference-to-diameter ratios larger than π. On a curved surface the ratio can also change with the circle’s size, which is how it reveals the curvature. The schoolbook definition quietly assumed a flat plane. In general relativity, spatial geometry near massive bodies can be curved as well, so a careful circumference-to-diameter measurement need not return π.
But π itself hasn’t changed. Its digits are not a property of local curvature. A half turn still measures π radians, and a full phase cycle still measures 2π.
So π is not merely 3.14. It measures a half turn, hiding in waves, orbits and repetitions, and its equality with circumference over diameter tells you something else: that the page is flat.
