Baudhāyana and the Śulba Sūtras: the Pythagorean Theorem in Ancient India

The Baudhāyana theorem is the rule, written in the Baudhāyana Śulba Sūtra, that the diagonal cord of a rectangle "produces" both areas that its two sides produce separately — in modern notation a² + b² = c². The text is an Indian manual for building fire altars, usually dated somewhere around 800–500 BCE, and it contains one of the oldest known general statements of the Pythagorean theorem.

The same text gives practical constructions, a list of whole-number right triangles and an excellent value for √2 — but no proof and no firm date, two facts that matter when people ask whether India knew the theorem "before Pythagoras".

What the Śulba Sūtras are

Śulba (also spelled śulva) means a cord or measuring rope, and a sūtra is a short rule meant to be memorized. The Śulba Sūtras are the geometry sections of larger ritual manuals attached to the Vedas. They tell priests how to lay out the platforms and brick altars used in Vedic fire sacrifices, using nothing more than a cord, a few pegs and careful rules.

Four main texts survive, each named after the teacher of the ritual school that preserved it: Baudhāyana, Āpastamba, Kātyāyana and Mānava. The Baudhāyana Śulba Sūtra belongs to the Taittirīya branch of the Black Yajurveda tradition. Nothing is known about Baudhāyana as a person beyond his name; we cannot even be sure that one individual wrote the whole text.

How old are they?

No early manuscripts exist. The texts were passed down orally for centuries, so historians date them indirectly, from their language and from which text seems to borrow from which. The results are ranges, not dates:

Some historians stretch the overall span much further — Carl Boyer gave the eighth century BCE to the second century CE. A fair summary is that the Śulba Sūtras were composed somewhere between about 800 and 200 BCE, with Baudhāyana's text near the early end of that range.

Baudhāyana's statement of the theorem

The key rule is a single line of Sanskrit. One common transliteration (spellings vary slightly between editions) reads:

dīrghacaturasrasyākṣṇayā rajjuḥ pārśvamānī tiryagmānī ca yat pṛthagbhūte kurutas tad ubhayaṃ karoti

George Thibaut, whose English translations in the 1870s introduced the Śulba Sūtras to Western scholars, rendered it as: "The diagonal of an oblong produces by itself both the areas which the two sides of the oblong produce separately." Word by word, it runs roughly like this:

"Produces" means "produces as a square": the area of the square built on a cord. So the rule says that the square on the diagonal equals the square on the length plus the square on the width. Because the diagonal cuts a rectangle into two right triangles, this is exactly the theorem for a right triangle with legs a and b and hypotenuse c.

Shortly before this, the text gives the special case for a square: the diagonal cord of a square produces twice its area. That is the rule a priest needs to double an altar while keeping its shape, and it is the same fact Plato would later dramatize in the Meno.

Two points deserve emphasis. First, the rule is stated for every rectangle, not just for a lucky set of numbers like 3, 4, 5, which is what makes it a general theorem rather than a recipe. Second, like the rest of the Śulba Sūtras, it is stated without a proof. The texts give rules to follow, not arguments for why they work; later Indian mathematicians, such as Bhāskara II in the twelfth century, supplied demonstrations.

Why altar builders needed it

Vedic ritual specified altars by shape and by area, and some rites required the same area to be produced in a different shape — a square turned into a circle, or an altar enlarged without changing its outline. You cannot do that with a cord and pegs unless you can build right angles reliably and add or subtract areas exactly. The diagonal rule does both jobs.

The most famous structure is the falcon-shaped altar (śyenaciti). In its standard form it was built from 1,000 fired bricks in five layers of 200, with a total area of 7½ square puruṣa (a puruṣa is the height of a man with raised arms, 120 aṅgula or "finger-breadths"). Each layer had to cover exactly that area with bricks of a few prescribed shapes, so the geometry had to be exact.

A simpler example shows the theorem at work. The great altar ground (mahāvedi) of the soma ritual is an isosceles trapezoid 30 units wide at the west, 24 units wide at the east and 36 units long from west to east (the unit is a pada or prakrama, a step).

To lay this out, the priests first fixed the east–west spine and then needed true right angles at both ends. Āpastamba's text does this with cords whose lengths are the three sides of a whole-number right triangle: peg the ends of one side, pull the cord taut at the marked point, and the corner is square. That technique relies on the converse of the Pythagorean theorem — the same principle behind the builder's 3-4-5 rule, and the one often (but without solid evidence) credited to the Egyptian rope stretchers.

Adding and subtracting squares

The Śulba Sūtras use the diagonal rule most often to combine areas. To make one square equal to two given squares, Baudhāyana's method is:

1. Mark a strip. From the larger square, cut off a rectangle whose width is the side of the smaller square.
2. Stretch the diagonal. The diagonal cord of that strip has, as its square, the large square plus the small one.
3. Build the new square on that cord. Its area is the sum of the two original areas.

Subtraction works the same way in reverse. To remove a small square from a large one, the text again cuts off a strip as wide as the small square, then swings a cord as long as the large square's side across the strip until it touches the far edge. The distance cut off along that edge is the side of the remaining square, because it is a leg of a right triangle whose hypotenuse is the large side: √b² − a². It is the same calculation you use to find a missing leg.

The texts also turn a rectangle into a square of equal area (exactly) and a square into a circle of nearly equal area. Baudhāyana's circle has a radius of half the side plus one third of the amount by which half the diagonal exceeds half the side; by modern arithmetic it is about 1.7% too large. The texts state such approximations as plainly as their exact rules, and historians disagree about whether the authors knew which was which.

The √2 rule: 1 + 1/3 + 1/(3·4) − 1/(3·4·34)

To double a square you need its diagonal, and to give the diagonal as a number you need √2. The Śulba Sūtras (the rule appears in Baudhāyana, Āpastamba and Kātyāyana) give it in words. In paraphrase: increase the measure by its third, increase that third by its own fourth, and take away the thirty-fourth part of that fourth. In symbols:

The true value is √2 = 1.4142136 (to seven decimal places), so the error is about 0.0000021 — roughly two millionths, correct to five decimal places. Each term of the rule improves on the one before:

Notice how the last correction works. The error of 17/12 is about 0.0024531, and the subtracted term 1 over 408 is about 0.0024510 — almost exactly the same size. The fractions also have a neat property: 17² = 289 is one more than 2 × 12² = 288, and 577² = 332,929 is one more than 2 × 408² = 332,928. Squares that are "twice a square, plus one" are what make these fractions such good approximations.

How the rule was found is not recorded. Bibhutibhushan Datta (1932) proposed a geometric reconstruction that cuts and rearranges pieces around a square; R. C. Gupta later suggested a method of successive corrections. Some historians have suggested a link with Mesopotamia, where the Old Babylonian tablet YBC 7289 (c. 1800–1600 BCE) gives √2 in base 60 as about 1.4142130 — even closer than 577/408. There is no direct evidence either way. For the reason no fraction can ever be exactly right, see why √2 is irrational.

Pythagorean triples in the texts

Right after the diagonal rule, Baudhāyana notes that it can be seen in rectangles with particular whole-number sides. His list, with the diagonals that the rule implies:

Āpastamba's text works with (3, 4, 5), (5, 12, 13), (8, 15, 17) and (12, 35, 37). Five of Baudhāyana's six are primitive triples, whose three numbers share no common factor; the sixth is the 5-12-13 triangle scaled by 3, which is the triangle hidden in the mahāvedi. The texts give no general formula for producing triples, unlike Euclid's formula, and no list as long as the Babylonian tablet Plimpton 322. To see where these fit among all the triples, browse the list of Pythagorean triples.

Was India first? Weighing the priority claims

In India the result is often called the Baudhāyana theorem (or Baudhayana theorem), and there is a solid case for that credit. The rule is general, clearly stated and, on most estimates, older than Pythagoras (c. 570–c. 495 BCE), who left no writings of his own; the stories linking him to the theorem come from authors writing centuries after his death. It is certainly older than Euclid's proof (c. 300 BCE).

Other evidence complicates any claim to be the first:

• Babylon is older. Old Babylonian tablets from about 1800–1600 BCE use the relationship in calculations and list large triples, roughly a thousand years before the usual dates for Baudhāyana.
• The dates are soft. With Baudhāyana placed anywhere from about 800 to 500 BCE, the text could be two centuries older than Pythagoras or roughly his contemporary.
• A rule is not a proof. The Śulba Sūtras state the theorem; the oldest surviving deductive proof is Greek (Euclid's), and an independent Chinese demonstration comes from the gougu tradition.

Historians also disagree about whether these traditions were connected. Abraham Seidenberg argued that geometry had a ritual origin and that either Babylon learned the theorem from India or both inherited it from an older common source; B. L. van der Waerden also favored a common source. David Pingree warned against treating the altar builders as "the unique origin of geometry," and Kim Plofker suggests that existing geometric knowledge may have been absorbed into ritual practice. No surviving evidence shows transmission in any direction.