Pythagoras Didn't Discover His Theorem

A clay tablet in the Yale Babylonian Collection carries four numbers and a tilted square, and if you do the arithmetic those numbers give you the length of the diagonal of a square to an accuracy that would embarrass most calculators built before electricity. The tablet is a student's exercise, scratched out roughly a thousand years before a man named Pythagoras was born on Samos. Whatever Pythagoras did in his life — and, as argued below, we know remarkably little of it for certain — he did not discover the relationship between the sides of a right triangle that bears his name. Someone in Mesopotamia already had it, generations before there was a Greece to have a mathematician in.

That is the easy part of this essay, and it is not really controversial among people who study the sources; it has simply not reached the textbook. The harder part is what to do with the gap between "knowing that a relationship holds in case after case" and "proving that it must hold in every case," because those are different achievements, and collapsing them is how the wrong man ends up famous. That distinction holds steady through three moves: the numbers predate him by a millennium, the man himself is almost invisible in reliable sources, and the proof we actually have is best read as the product of a tradition rather than a person — with the name attached later, the way a citation gets attached to a paper, not the way a signature gets attached to a deed.

A tablet that already knew

Start with the artifacts, because they are the least contestable evidence in this whole story. Plimpton 322, now in Columbia University's collection (the object was purchased by the publisher George Arthur Plimpton in the 1920s and bequeathed with his library), is a broken Old Babylonian tablet, probably from around 1800 BCE, ruled into columns of numbers in the Babylonians' base-60 notation.¹ Read as modern mathematicians read it, several of its rows are what we would call Pythagorean triples — sets of three whole numbers where the square of the largest equals the sum of the squares of the other two, the numerical signature of a right triangle. Otto Neugebauer and Abraham Sachs, who first published a full edition of the tablet in the 1940s, worked out that the numbers could plausibly have been generated from a method equivalent to a general algebraic identity for producing such triples, which would mean whoever made this list understood the relationship as a pattern to be manufactured, not just stumbled into by measuring a few sturdy walls.¹

What the tablet was for is genuinely disputed, and the case is not closed. Eleanor Robson has argued, against a popular reading that treats Plimpton 322 as the world's oldest trigonometric table, that it is much more plausibly a teacher's aid — a list generated from reciprocal pairs, of the kind Babylonian scribal schools used constantly, that happens also to be a list of triples, produced for setting problems rather than for measuring angles.² A 2017 paper reasserted the trigonometry reading in a stronger form; Robson's objections still stand, and a live argument among cuneiform specialists is not adjudicated here. What both sides agree on, and what matters here, is that the scribe who made this tablet already possessed, and could generate at will, sets of numbers satisfying what we call the Pythagorean relation — table or no table.²

The companion piece is smaller and less argued over. YBC 7289, also at Yale, is a round tablet showing a square with both diagonals drawn, marked with a side length of 30 and, along one diagonal, two numbers in sexagesimal: 1;24,51,10 and 42;25,35.¹ The first number is an approximation of the square root of two accurate to about six decimal digits; the second is that approximation multiplied by the side length, giving the diagonal directly. This is a school tablet doing exactly the geometric operation the later theorem formalizes — squaring a side, taking a square root, relating it to a diagonal — worked by hand, in a base-60 place-value system that a different essay on this site traces all the way into the clock on your wrist.¹ Why Your Watch Is Babylonian is about the survival of that number system; this essay is about what the scribes did with it. Whoever wants to see these tablets in person, or in careful photographic surrogate, can start with the museum and library holdings mapped out in Where the Tablets Are.

Rope, altars, and the line between knowing and proving

The Babylonian material is the strongest case because it is dated cuneiform, physically excavated and physically legible. Two other traditions are usually invoked alongside it, and both deserve to be stated more carefully than they usually are.

The Egyptian claim, repeated in almost every popular account of the theorem, is that Egyptian surveyors used a rope with twelve equally spaced knots, stretched into a 3-4-5 triangle, to lay out a right angle for field boundaries and temple foundations after the Nile's flood erased them each year. The image is vivid, and the underlying geometric fact — that a 3-4-5 triangle is a right triangle — is real. But the direct evidence is thinner than the story deserves: it traces back to a later Greek source, a fragment attributed to Democritus mentioning Egyptian "rope-stretchers," and to classical writers calling Egyptian surveyors harpedonaptai, rope-fasteners.³ No surviving Egyptian mathematical papyrus — not the Rhind, not the Moscow — states or uses the general right-triangle relation; the papyri show scribes highly capable at practical geometry, but the 3-4-5 rope method is an inference from later Greek testimony, not a native Egyptian text. It is plausible surveying practice, and plausible practices often go unrecorded, but it belongs here as a hedge, not a second Babylon.

India's case is more textual and more explicit, though its dating is its own contested question. The Śulba-sūtras, a set of ritual manuals attached to the Vedic tradition and concerned with the precise geometric construction of fire altars, contain statements that are unambiguously the general relationship: the diagonal of a rectangle, expressed as a rule for the area of the square on it, related to the squares on the two sides. The Baudhāyana Śulba-sūtra states a version of this rule explicitly, as a method for constructing altars whose area matched a required specification exactly.⁴ Dating these texts is difficult because they survive through a manuscript and recitation tradition rather than as fixed archaeological objects; scholars place the core content of the Śulba-sūtras across a wide range, roughly the later centuries of the second millennium BCE down through the middle of the first millennium BCE, with most estimates for Baudhāyana's text clustering well before Pythagoras's traditional lifetime.⁴ Kim Plofker's history of Indian mathematics treats the Śulba-sūtra rule as a genuine ritual-geometric statement of the relationship, independent of any Greek or Babylonian transmission, developed for the specific practical problem of altar construction rather than as abstract theory.⁴

So: three traditions, one solidly dated and numerically explicit, one plausible but thinly attested, one textually explicit but harder to date precisely — and every one of them earlier, or at minimum no later, than the man Greek tradition credits. Here is the distinction promised above, because it is where this essay actually turns and not where it hides. A scribe who can generate triples, or approximate a diagonal to six digits, or state a rule for altar construction has demonstrated that the relationship holds in case after case, and has a working method for producing more cases. None of that is the same act as showing, from a small set of agreed starting assumptions, that the relationship must hold for every right triangle whatsoever, with no possible counterexample — a general deductive proof, the kind that starts "let ABC be a right triangle" and ends "which was to be demonstrated." The Babylonians, the Egyptians (on the thin evidence available), and the authors of the Śulba-sūtras had powerful, correct knowledge of instances and of methods for generating them. None of that evidence is a text that argues the general case from axioms. That is a real distinction, not a technicality invented to protect Greek originality: the relationship is not a Greek discovery, and the deductive proof, wherever it first appears, is a different kind of object from a list of confirmed instances.

Knowing that a relationship holds in case after case is not the same act as proving it must hold in every case, and the wrong man gets famous exactly at the point where that distinction goes unstated.

The man before the hagiography

Now the second move, which is less about tablets than about how thin the historical Pythagoras actually is once you clear away everything written centuries after he died.

Pythagoras left no writings that survive, and by most assessments of the ancient testimony, he probably left none at all. He was born on Samos sometime in the sixth century BCE, migrated to Croton in southern Italy, and founded there a community that historians generally describe as at once religious, ascetic, and political — bound by dietary restrictions (the famous, oddly specific rule against eating beans), rules about burial and ritual purity, and a doctrine of the transmigration of souls.⁵ The earliest substantial mention of him in surviving Greek literature is Herodotus, writing roughly a century after Pythagoras's death, and Herodotus's Pythagoras is not a geometer at all — he appears in a digression about a Thracian figure named Salmoxis, credited (skeptically, by Herodotus's own account) with importing a Greek-style doctrine of the soul's immortality, and Herodotus calls him, in paraphrase, not the weakest sophist among the Greeks.⁶ That is the whole of it: a religious teacher, described secondhand, a century out, with no triangle in sight.

The mathematical Pythagoras — founder of a school that discovered the theorem, worked out the harmonic ratios of the musical scale, and organized geometry as a deductive discipline — belongs almost entirely to sources written five to nine centuries after his death: Diogenes Laertius's biographical compendium in the third century CE, and above all Porphyry and Iamblichus, Neoplatonist philosophers of the third and fourth centuries CE writing devotional biographies meant to present Pythagoras as a semi-divine sage, a model for their own school's spiritual practice.⁷ These are rich, entertaining sources, and they are also, on any ordinary standard of evidence, the wrong kind of source for establishing what a sixth-century Greek actually did: hagiography for a live religious-philosophical movement that needed a founder-hero, written long after the fact, drawing on earlier material that is itself largely lost and cannot be checked.

Walter Burkert made the most influential case for taking this seriously as a historical problem rather than an incidental gap. In Lore and Science in Ancient Pythagoreanism (the English translation, 1972, of his 1962 German study), Burkert argued that the image of Pythagoras as mathematician and natural philosopher is largely a later construction, built up by attributing to the founder discoveries that actually belonged to fifth-century Pythagoreans like Philolaus, and that the early, more reliable evidence points instead to a figure closer to a religious and shamanistic teacher concerned with the soul, ritual purity, and a way of life, not with axiomatic geometry.⁸ Carl Huffman's detailed philological work on Philolaus — whose fragments are the earliest substantial Pythagorean writing to survive in any form — supports the general caution: Philolaus's own concerns are cosmological and numerological, organized around number as the structuring principle of reality, and Huffman is careful not to read a rigorous mathematical program back into him, let alone back further onto Pythagoras himself.⁹

Burkert's minimalism is not the last word. Leonid Zhmud, in Pythagoras and the Early Pythagoreans, has argued at length against Burkert's reconstruction, contending that the ancient sources, read with a different and in his view more generous method, support real Pythagorean involvement in mathematics and natural science well before Philolaus, and that Burkert's skepticism swings too far toward treating Pythagoras as purely a religious figure.¹⁰ This is a live disagreement between serious historians of ancient philosophy, not a settled question. What both sides agree on, even where they disagree on how far the pendulum swings, is that the evidence closest to Pythagoras himself is thin and hard to pin to specific mathematical claims — and that the richly circumstantial version most people know, brotherhood and bean-taboo and secret geometric proofs, comes from sources many centuries removed.

The proof, the ox, and a Roman's doubt

Which brings the third strand into view: even the specific attribution of the proof — not just general mathematical brilliance, but this particular deductive demonstration — is late, and does not rest on any text contemporary with Pythagoras or even close to it.

The proof we actually have, the one you can still read today essentially unchanged, is Euclid's, in Book I, Proposition 47 of the Elements, compiled in Alexandria around 300 BCE — two and a half centuries after Pythagoras.¹¹ Euclid's proof proceeds from his own definitions, postulates, and earlier propositions about parallel lines and areas, building squares on each side of a right triangle and showing, through a construction of auxiliary lines that generations of students have nicknamed the "windmill" or the "bride's chair," that the square on the hypotenuse has exactly the combined area of the squares on the other two sides. It is a general argument. It works for every right triangle, not a sample of them, and it works because of the logical structure Euclid built beneath it — the deductive apparatus that is really the achievement, since the theorem itself was, as established above, already known as a numerical fact a millennium earlier in Mesopotamia. Euclid does not attribute Proposition 47 to Pythagoras. He does not attribute it to anyone. It simply appears, numbered, in its place in the argument.

The attribution to Pythagoras by name comes from later commentators, most influentially Proclus, writing his commentary on Euclid's first book in the fifth century CE — roughly a thousand years after Pythagoras and seven or eight centuries after Euclid.¹¹ Proclus, drawing on an earlier history of geometry by Eudemus that does not itself survive, reports the tradition that Pythagoras discovered the theorem and, according to a story that circulated with the attribution, sacrificed an ox to the gods in gratitude. The ox-sacrifice anecdote is almost certainly legendary on its own internal evidence: several ancient sources report the Pythagorean community as vegetarian, or at minimum committed to a doctrine of the kinship of all souls that made animal sacrifice deeply problematic, and Cicero, writing in the first century BCE, explicitly doubts the ox story for exactly this reason — a Pythagorean, on the account of Pythagorean practice everyone already had, should not have been offering blood sacrifice at all.¹¹

A legend that contradicts the reported character of its own subject is a legend explaining itself away.

Whoever writes the history gets to name the theorem

Put the three moves together and a more defensible picture comes into focus. The relationship between the sides of a right triangle was known and exploited across at least three ancient cultures long before there was a datable Greek mathematician to credit it to. A rigorous, general, axiom-based proof of that relationship is a real and separate achievement, and it does belong to the Greek deductive tradition — visible, unmistakably, in Euclid's Elements, built out of definitions, postulates, and a chain of earlier propositions, none of which required a single founding genius to exist. And the specific act of pinning that achievement to one man, with a name, a brotherhood, and an ox, is a later Greek and Roman habit of historiography, arriving centuries after both the man and the proof, in sources — Proclus drawing on the lost Eudemus, Porphyry and Iamblichus writing devotional biography — that were doing something other than sober history when they wrote it down.

The most useful way to describe what "the Pythagorean theorem" names is a citation practice rather than a historical claim: a convention, fixed early and never seriously revisited, for pointing at a piece of mathematics by attaching it to the most famous available name in the tradition that formalized it, whether or not that name did the formalizing. This is not a uniquely Greek habit, and it is not unique to mathematics — it is close to the same mechanism traced in Greek Myth Was Never a Religion, where a body of stories with no scripture and no central author still got treated, retroactively, as if it must have had one; here a body of deductive practice with no single inventor got treated as if it must have had one too, and the founder-figure of an already-famous religious school was the obvious candidate to receive the credit once someone went looking for a name to attach. Credit, in other words, tends to accrete to whoever is already famous when the bookkeeping happens, and the bookkeeping in this case happened seven or eight hundred years after the fact, by writers with their own reasons for wanting a single luminous founder at the head of the tradition.

None of this requires believing Pythagoras invented nothing, or that he did not exist, or that the deductive turn in Greek geometry was somehow borrowed rather than developed. None of that is argued here, and the evidence does not support it. Greek mathematics, whatever its debts to Babylonian numerical technique — and those debts are real and worth taking seriously — developed something none of the surviving Babylonian, Egyptian, or Indian material shows: an explicit, self-conscious insistence on proving general claims from stated first principles, visible across Euclid's entire structure and not confined to one proposition. That is a genuine, documented achievement of the Greek mathematical tradition taken as a whole. What is not supported by the evidence is the narrower claim that one specific, nameable person performed it, and that we know this because a devotional biographer said so a thousand years later, citing a historian whose own book we no longer have. The theorem is real. The proof tradition behind it is real and worth admiring. The name on it is a label applied long after the fact, by people who needed a hero and had one already sitting in the historical record, waiting only to be handed a triangle he probably never touched.