Babylon Did Algebra for a Thousand Years Without Inventing Algebra

Open BM 13901, a well-known collection of Old Babylonian problem texts, and you will find a scribe solving what we would now write as a quadratic equation — and not writing anything we would recognize as one.¹ No x, no equals sign, no general formula. Instead: "I have added the surface and my side, it is three-quarters," followed by a sequence of instructions — take half of this, square that, add this other thing, take the square root — worked all the way through to a specific numerical answer.¹ The method is exactly the one taught in a modern algebra class for completing the square. The scribe who wrote it had never heard of algebra, had no notation for it, and lived more than a thousand years before anyone would coin the word. This essay is about that gap, and about why it matters more than it looks like it should.

A procedure, not an equation

Old Babylonian mathematics belongs to the early second millennium BCE, the flourishing period of the scribal schools of southern Mesopotamia, and its mathematical tablets are one of the best-documented technical literatures to survive from anywhere in the ancient world.² Among them are two famous, very different kinds of object. YBC 7289 is a small round tablet, probably a student's hand exercise, that gives a value for the diagonal of a unit square accurate to within a fraction of a millionth — a sexagesimal approximation of what we call the square root of two, computed to a precision that has nothing to apologize for next to a modern calculator.³ Plimpton 322, a tablet of number pairs whose exact purpose is still argued over, has already had its own treatment on this site and is not re-litigated here.⁴ BM 13901 belongs to a third category: a collection of some two dozen problems, mostly involving a square's side and its area, worked out step by step, in words, toward specific numbers.¹

What these problem texts have in common, and what makes them worth taking seriously as mathematics rather than as curiosities, is that the scribes were not merely arithmetic clerks. They were solving problems of a type we would classify as quadratic — a length and an area combined in a single relationship, an unknown quantity constrained by more than one condition — using a reliable, repeatable procedure that gets the right answer every time it's applied.⁵ That is genuine mathematical competence, and calling it "not really algebra" because it lacks notation risks sounding like a modern chauvinism, a way of disqualifying an achievement because it wasn't dressed the way we dress our achievements. That mistake is worth avoiding. But there is a difference worth taking seriously that isn't about intelligence at all; it is about form: every one of these procedures is tied to specific numbers. There is no version of the text that states the method as something you could apply to an unknown length and an unknown area in the abstract, independent of which numbers you happened to plug in. The scribe solves "this problem, with these numbers," beautifully, and then the next problem, with different numbers, gets solved again from the start.

This is worth sitting with, because it is easy to read a worked example and mentally supply the generalization the text itself never states. When BM 13901's first problem walks through "the surface and my side is three-quarters" and arrives at the right side-length, a reader trained in modern algebra automatically translates the steps into x² + x = 3/4, solves it symbolically in their head, and then credits the tablet with having done the same thing. But the tablet doesn't do that. It does the arithmetic, on those particular numbers, and stops. There's no line anywhere in the Old Babylonian corpus that says, in effect, "for any surface-plus-side problem, do the following" — no case where the specific numbers are replaced by a placeholder and the procedure is stated once, for every problem of the type. The tablets contain worked instances. They don't contain the class.⁵

Cutting and pasting a square

How were the scribes actually thinking about these problems? This is where the historiography gets genuinely contested, and it's one of the more interesting arguments in the field precisely because it turns on how you read a text that gives you steps but not reasons. The older influential view, associated above all with Otto Neugebauer's foundational work on Babylonian mathematics in the mid-twentieth century, essentially read the procedures as algebra by another name: the scribes were manipulating quantities the way we would manipulate an unknown in an equation, and the specific numbers were incidental to an underlying symbolic-style operation that the notation simply couldn't display.⁶ On this reading, the Babylonians had algebra; they just lacked the alphabet to write it in.

The reading that has become dominant since, and the more persuasive one, comes from Jens Høyrup's long philological study of the vocabulary these texts actually use.¹ Høyrup went back to the specific Akkadian terms for the operations — the words that get translated blandly as "add" or "multiply" — and argued that several of them carry a much more concrete, spatial meaning than a modern algebraic gloss suggests: not abstract addition but literally joining two areas together, not abstract multiplication but constructing a rectangle from two given lengths. On this reading, "completing the square" wasn't a metaphor borrowed later by Renaissance mathematicians to explain an abstract technique to students — it was, for the Babylonian scribe, actually and literally cutting a piece off one part of a figure and pasting it onto another, all worked out on a diagram you could in principle have drawn in the dirt, with a real rectangle and a real strip and a real missing corner. The "unknown side" is a real, if unmeasured, length of a real geometric figure, not a variable standing for a number in an equation.

This is a genuine disagreement among people who read the cuneiform, not a settled consensus. Høyrup's geometric interpretation has become the more widely cited reading in recent decades, largely because it makes better sense of the specific vocabulary and the order of operations than a straightforwardly algebraic gloss does, but it has not eliminated the older view so much as displaced it as the default, and there is room to argue about how far the geometric reading should be pushed for every problem in the corpus.¹ Eleanor Robson's social history of Mesopotamian mathematics treats the cut-and-paste geometric reading as the more defensible one while stressing, rightly, that the scribal-school context — these were, after all, training texts, produced inside an institution set up to reproduce a caste of professional administrators — matters as much as the abstract mathematical content for understanding why the problems look the way they do.²

Either way — concrete cut-and-paste geometry, or an algebra-shaped procedure trapped in numerical clothing — the outer fact is the same, and it's the fact that carries this essay's argument. Whatever the scribes were picturing in their heads, what they wrote down was always a specific, numerically worked case. Nothing in either camp's reading of BM 13901 produces a Babylonian text that states a general rule independent of particular numbers. That absence is real under both interpretations, and it's the thing worth explaining.

What a worked example can't do

Here is the argument the tablets are actually making, once you stop asking whether the Babylonians were "doing algebra" and start asking what a worked numerical procedure can and can't do compared to a stated general rule. A worked example, however elegant, is bound to its instance. It shows you that this problem, with these numbers, comes out this way, following these steps. To apply it to a different problem, you either have to already understand what's invariant across the two cases — in which case you've done the generalizing work yourself, silently, in your own head — or you have to be shown a new worked example for the new numbers, redone from scratch by someone who already knows the method.

A stated general rule does something categorically different. It doesn't just solve a problem; it names the type of problem being solved, and it does so in a form that can be checked, taught to a stranger who has never seen an instance of it, extended to cases the original author never worked through, and built into something larger without anyone having to re-derive it from first principles each time. This is not a claim about which method requires more intelligence — the Babylonian scribes were plainly capable of exactly the reasoning a general rule would require, since they applied it correctly, problem after problem, for centuries. It's a claim about portability. A technique nobody can express compactly does not compound. It has to be re-taught, case by case, generation by generation, and every re-teaching is a chance for the thread to break.

A worked example, however elegant, is bound to its instance. A stated general rule names the type of problem being solved — and a named type is something a stranger can pick up, check, and build on.

What was missing in Babylon, on this account, was not mathematical insight. The scribes had that in abundance; nobody produces a correct, general-purpose quadratic-solving procedure by accident, and nobody keeps applying it correctly across dozens of numerically distinct problems without understanding, at some level, why it works. What was missing was a compression format — a way of writing the method down that separated the invariant procedure from the particular numbers it happened to be demonstrated on. Without that separation, every problem is its own small monument, complete and correct and entirely local.

Al-jabr gives it a name, still no symbols

The word itself, and the first systematic treatment of the type of problem rather than the instance, comes from somewhere else entirely: ninth-century CE Baghdad, more than two thousand years after BM 13901 was inscribed, in a treatise by Muhammad ibn Musa al-Khwarizmi usually dated to the 820s.⁷ Al-Khwarizmi's book takes its later Latin name, and our word "algebra," from one of its two central operations, al-jabr, restoring a diminished quantity by adding an equal amount to both sides of a relationship — paired with al-muqabala, balancing or comparing like terms. What al-Khwarizmi did that no Babylonian text does is classify: he laid out the different types of equation that reduce to combinations of squares, roots, and numbers, gave a named, general method for each type, and proved geometrically why each method works, independent of any one set of numbers.⁸ A reader of al-Khwarizmi's treatise comes away knowing how to solve every problem of a given type, not just the one worked example in front of them. That is the systematic move the Babylonian material never makes, whichever way you read Høyrup's geometry.

It would be a mistake, though, to read this as the arrival of algebra in anything like its modern form, and it's worth being precise about what al-Khwarizmi still didn't have. His treatise is written entirely in words. There is no symbol for the unknown, no equals sign, no notation for a coefficient — a squared quantity is "the square" (mal), the unknown itself is "the root" (jidhr), and every equation is stated and solved as a sentence, in the same rhetorical style the Babylonian scribes used, just applied now to the general case instead of the particular one.⁸ Roshdi Rashed's work on the broader Arabic mathematical tradition that grew out of al-Khwarizmi's treatise — through mathematicians like al-Karaji and, later, Omar Khayyam working on cubic equations — shows a tradition that became steadily more systematic and more comfortable treating algebra as a subject in its own right, still without ever adopting a symbolic notation for it.⁹ The rhetorical style survives for centuries after the word "algebra" exists. Having a name for the discipline and having a compact notation for writing it turn out to be two separate inventions, arriving centuries apart.

Even the man the word "algebra" is named for wrote every equation as a sentence. The name arrived a full symbolic revolution before the symbols did.

The x arrives later, and the compression closes

Symbolic algebra — letters standing for known and unknown quantities, manipulated according to fixed rules, independent of any particular numerical instance — is a distinctly later, European development, and even here the story runs in stages rather than a single leap. The French mathematician François Viète, in a work usually dated to 1591, introduced the systematic use of letters for both knowns and unknowns, conventionally consonants for the one and vowels for the other, which let him write a general equation as a genuine algebraic expression rather than a sentence about a specific number.¹⁰ That convention wasn't yet the one we use today, and it took another mathematician, several decades later, to settle something close to the modern notation: René Descartes, in the appendix to his 1637 Discourse on Method known as La Géométrie, used letters from the end of the alphabet — x, y, z — for unknowns and letters from the beginning for knowns, the convention that has stuck, essentially unchanged, into every algebra classroom since.¹¹ Victor Katz's survey history of mathematics traces this same sequence — Babylonian procedure, then al-Khwarizmi's rhetorical but systematic classification, then centuries of further rhetorical and syncopated algebra across the Islamic world, Indian mathematics, and medieval Europe, before Viète's and then Descartes' notation finally arrives — as a genuinely long relay, not a single breakthrough with a single hero.⁸

Laid end to end, the sequence is genuinely strange when you say it plainly: the mathematics comes first, worked correctly, in Babylon, in the early second millennium BCE. The word, and the first general treatment of the type, comes roughly two thousand years later, in Baghdad, in a different language and a different civilization entirely. The symbol — the actual x — comes later still, another seven or eight centuries after that, in France. Three separate inventions, three separate cultures, roughly three thousand years apart at the extremes, and none of them redundant with the others. Knowing how to solve the problem, having a name and a systematic classification for the type of problem, and having a compact notation to write the classification in are three different achievements, and history delivered them one at a time, centuries apart, to three different civilizations that had almost nothing else in common.

Writing itself, in this same region, was invented for records — tallies of grain and sheep and debts owed, not for literature, which arrived centuries later riding on infrastructure built for bookkeeping.¹² Mathematics inherited that same institutional shape. The scribal schools that produced BM 13901 were training administrators, not researchers in pure mathematics for its own sake, and the problem texts they copied and worked through read like exactly what an accounting-descended discipline would produce: itemized, worked, correct, and stubbornly tied to the specific case in front of the student, because the point of the exercise was competence at solving problems that would actually come up, not the discovery of a compact, general statement of a mathematical type. A mathematics born from record-keeping stays a mathematics of worked examples for as long as nobody needs it to be anything else — and for over a thousand years, in Babylon, nobody did.

The base-60 arithmetic that survives on your wrist today shows how durable a Babylonian convention can be once it gets embedded in the right infrastructure, and the algebra story is almost the mirror image of that one. Sexagesimal notation won by being compact and easy to divide, and it rode that portability all the way to the present. The Babylonian algebraic procedure had no equivalent portability, not because the underlying mathematics was weak, but because it had never been separated from its numbers. It couldn't be copied into a new context the way a compact rule can, and so, as far as the surviving record shows, it stayed exactly where it was invented, re-derived and re-taught inside one scribal tradition for as long as that tradition lasted, never generalized into the kind of statement a stranger centuries later could pick up and extend.⁵ Compare that to what happened to al-Khwarizmi's classification once Latin translations of Arabic algebraic texts began circulating in medieval Europe: a compact, general statement of a method is exactly the kind of object that survives being carried into a new language, a new century, and a new set of hands, because there's a stable, checkable core to translate rather than a pile of worked instances to somehow infer a rule from.

None of this is a story about Babylonian mathematics being primitive; the arithmetic embedded in BM 13901 and YBC 7289 is precise, reliable, and in places startlingly sophisticated, and the sheer bulk of the surviving tablet record makes clear this was a living, working discipline practiced at real scale, not an isolated curiosity.⁴ It's a story about what a civilization can and can't do with a correct method once it has no way to say the method out loud, apart from the numbers it was demonstrated on. The Babylonians solved the problem. Al-Khwarizmi named the type of problem and classified its solutions, still in words. Viète and Descartes gave it a notation compact enough to travel, prove things about, and build on without anyone having to walk through a worked numerical instance first. Three different achievements, arriving three thousand years apart, because knowing how to do something and being able to say what you did, in a form someone else can pick up and extend, turn out to be two entirely different inventions — and the second one, on the evidence of how long the first one sat waiting for it, is very far from automatic.