The Pythagorean relationship

In brief

A carpenter checks that a wall is truly square not with a protractor but by measuring three, four, and five units along the two walls and the diagonal; if the diagonal comes out exactly five, the corner is a genuine right angle. Why should measuring three particular lengths certify an angle at all? The answer is one of the oldest and sturdiest facts in mathematics: in any right triangle, the square built on the longest side always has exactly the same area as the two squares built on the other two sides put together. That fact, discovered independently by more than one ancient culture and proved rigorously by the Greeks, is the hinge between shape and number: it lets you compute an unreachable distance, a ladder's reach, a diagonal brace, a line of sight, from two measurements you can actually take.

The full treatment

First look: squares on the sides

Draw a right triangle with legs of length three units and four units, then draw a square directly on each of the three sides. Count the unit squares packed into each drawn square: the square on the leg of length three holds nine unit squares, the square on the leg of length four holds sixteen, and the square on the hypotenuse, the side opposite the right angle, holds twenty five. Nine plus sixteen is twenty five. That is not a coincidence of this one triangle: the same equality, area of the square on one leg plus area of the square on the other leg equals area of the square on the hypotenuse, holds for every right triangle, whatever its actual lengths. Naming the two legs a and b and the hypotenuse c, the relationship is written a^2 + b^2 = c^2, where a^2 means the area of a square whose side is a.

Building the idea: what makes a triangle right

Before deriving anything, fix what is being assumed. A right angle is a quarter turn, the angle a square corner makes, and a right triangle is any triangle containing exactly one such angle. The side opposite the right angle is always the longest side, called the hypotenuse; the other two sides are the legs. The claim links three lengths purely through the presence of that one right angle; it says nothing about triangles with no right angle, and indeed the plain relationship fails for those, a fact worth holding onto for later.

The derivation: rearranging four triangles into two squares

Here is a proof that needs no trust beyond cutting and rearranging, and no arithmetic beyond expanding a product. Take four identical copies of the right triangle, legs a and b, hypotenuse c. Arrange them inside a square whose side length is a plus b, placing one triangle in each corner, each rotated a quarter turn from its neighbor, so that the four hypotenuses form a smaller, tilted square in the middle. Two things are now true about the area of this big square, side length a plus b, so its total area is a plus b, times a plus b.

First way to compute that same area: it is the four triangles plus the tilted inner square. Each triangle has area one half times a times b, so four of them total two times a times b. The inner square has side c, so its area is c squared. Total: two a b plus c squared.

Second way to compute the very same area, since it is the same square: expand a plus b, times a plus b, directly as a times a, plus a times b, plus b times a, plus b times b, that is a squared plus two a b plus b squared.

Both expressions describe the identical square, so they must be equal: a squared plus two a b plus b squared equals two a b plus c squared. The term two a b appears on both sides and can be canceled, an actual arithmetic step, not an assumption, leaving a squared plus b squared equals c squared. That is the whole relationship, produced by nothing more than rearranging four copies of one triangle inside one square and counting its area two different ways. The right angle entered only once, in the claim that the four rotated triangles leave a genuine square, rather than a slanted rhombus, in the middle; that the inner shape truly is a square, all sides equal length c and all corner angles square, follows because each of its angles is what remains of a straight edge after removing the triangle's other two angles, and those two angles sum to a right angle precisely because the triangle's third angle is a right angle. Change that one right angle to something else and the inner figure stops being a square, which is exactly why the whole relationship is special to right triangles.

Reading the formula

a^2 + b^2 = c^2 is not a rule to memorize; it is a compressed statement of an area equality, provable by dissection, that holds only when the angle between the two legs a and b is a right angle. Given any two of the three lengths, and knowing the triangle is right, the third is fixed: c equals the square root of a squared plus b squared, or, solving for a leg, a equals the square root of c squared minus b squared.

Lineage

Right triangles with the three-four-five property were used for accurate right angle construction well before any proof existed: Old Babylonian tablets from around 1800 BCE, including the tablet known as Plimpton 322, list Pythagorean triples, sets of whole numbers satisfying the relationship, suggesting working, computational knowledge centuries before Pythagoras. Ancient Egyptian and Chinese sources describe similar practical uses, and a version of the area argument appears in the Chinese text Zhoubi Suanjing. The Greek contribution, credited traditionally to Pythagoras and his school in the sixth century BCE though no writing of his survives, was to demand and supply a general proof rather than a catalog of working examples. Euclid's Elements, Book I, Proposition 47, gives a different, equally rigorous proof by comparing triangle areas rather than rearranging squares, preserved in Heath's translation; the rearrangement proof used above is one of hundreds of independent proofs collected over the centuries, including one attributed to a sitting United States president.

The strongest case for it

The relationship is not an approximation: given an exact right angle and exact leg lengths, the hypotenuse it predicts is exact, and this has been checked by proof, not merely by measurement, for over two thousand years. Its reach is enormous because a right angle is the single most common reference angle in built and natural structures: it lets a surveyor find the width of a river by measuring only along one bank and one perpendicular sightline, and lets a carpenter square a foundation with a tape measure and no angle tool. It also underlies the reverse move, computing straight line distance between two points from their horizontal and vertical separation, the basis of the distance formula used throughout coordinate geometry, physics, and navigation. Because it is a statement about area, it generalizes cleanly: the same equality holds if the three squares are replaced by any three similar shapes, semicircles or similar triangles, built on the three sides, so long as the same shape is used on each side, a theorem Euclid proves immediately after the square case.

The strongest case against it

The relationship is tied absolutely to one right angle; for any other angle between the two sides, a squared plus b squared is either more or less than c squared, not equal to it, and the correction term involves the cosine of that angle, a more general result called the law of cosines. A common misconception is to apply a^2 + b^2 = c^2 to any triangle with three known sides; without a confirmed right angle the formula gives a wrong answer, silently, with no warning that it does not apply. Second, the proof lives entirely in flat, Euclidean space; on a curved surface, the surface of a sphere for instance, a triangle can have a genuine right angle and yet the three sides fail the relationship altogether, because the area argument depends on flat squares tiling a flat plane, an assumption that stops holding on curved ground. Third, real measurement is never exact, so applying the relationship to a physically measured triangle, a plot of land or a building corner, inherits whatever error was present in the original length measurements, and that error is magnified through the squaring.

Where it stands now

The relationship is proved, not merely observed, and has been for over two thousand years by multiple independent methods; there is no live mathematical dispute about its truth within flat, Euclidean geometry. Its status as a special case of the more general law of cosines, and its failure on curved surfaces, are both equally well established, sharp known boundaries rather than open questions. What continues to be active is not whether it is true but how it generalizes, into higher dimensions, where an analogous sum of squares relationship holds for right angled boxes, and into non Euclidean geometries, where it is replaced by trigonometric analogues.

Test yourself

A tall pole stands perfectly vertical, its base fixed to level ground, and you are not allowed to climb it or use any measuring tool that reads angles. Using only a tape measure and the relationship developed above, describe a procedure to find the pole's exact height by working entirely on the ground, then explain why your procedure requires the base of the pole to make a right angle with the ground, and what would go wrong, specifically, if the pole leaned even slightly off vertical.