Arabic Numerals Are Indian, and the Arabs Say So

Every digit on this page — 0 through 9, the ones doing the actual work of counting — was invented in India, and nobody anywhere in the long chain that carried them to your screen has ever seriously disputed it. Not the Baghdad mathematician who wrote the first surviving treatise on how to calculate with them. Not the Pisan merchant's son who brought them into Europe. Not, notably, Arabic itself, which to this day calls them al-arqām al-hindiyya — the Indian numerals. The dispute, such as it is, belongs entirely to English, French, and the rest of the languages that call them "Arabic numerals" and mean, without quite meaning to, the numerals we got from the Arabs. That phrase is not wrong. It is just naming the wrong link in the chain, and the reason it does that turns out to be a rule general enough to explain a Thanksgiving bird, a steel ingot, and a counting system carved in stone by the Romans.

The part nobody disputes

Start with what the numerals actually are, because the achievement gets flattened if you only ever meet it as a symbol set. A decimal place-value system needs two things working together: digits whose meaning changes with position — a 3 worth three in one column and thirty in the next — and a way to mark an empty column so the positions don't collapse into each other. Babylonian scribes had a version of the first idea more than a thousand years earlier, in base 60, and eventually a placeholder mark for the second, but not, for most of that system's working life, a zero you could compute with — a subtlety traced elsewhere on this site in Why Your Watch Is Babylonian. What emerged in India was different in kind: a base-ten place-value system with zero treated as a number in its own right, subject to its own arithmetic, not merely a gap-filler.

The clearest surviving statement of that arithmetic is the astronomer Brahmagupta's Brāhmasphuṭasiddhānta, composed in 628 CE. In it Brahmagupta lays out rules for adding, subtracting, multiplying, and dividing with zero and with negative quantities — a fortune and a debt, in his own framing — treating them as ordinary objects of calculation rather than as special cases to be avoided. Some of his rules are shaky by modern standards; famously he treats division by zero as yielding zero, which is not the answer later mathematics settled on. But the shakiness is itself evidence of the ambition: nobody bothers writing rules for dividing by zero unless zero has already stopped being a mere placeholder and started being a number that arithmetic has to have an opinion about. Kim Plofker's history of Indian mathematics places Brahmagupta's work in a longer indigenous tradition of place-value calculation and numeral forms stretching back through earlier Indian inscriptions and manuscripts, not as a single isolated leap.¹ The Sanskrit numeral glyphs themselves — the actual shapes, ancestral to both the Indian scripts' digits and, eventually, to ours — descend from still older Brahmi numeral forms used across the subcontinent in the centuries before Brahmagupta, refined rather than invented by the astronomers who wrote the arithmetic down.²

What travelled out of India was therefore not a set of pretty symbols. It was a working method: any number, however large, written with ten reusable marks whose value depends on where they sit, with a tenth mark for nothing at all, executable by a person who has memorized ten shapes and a handful of column rules rather than a scribe who has memorized a different symbol for every power of ten. That is the entire payload. Everything else in this essay is the story of what happened to it on the way to your keyboard.

Al-Khwārizmī named his own source

The next fixed point in the chain is Baghdad, under the Abbasid caliphate, in the early ninth century, where the mathematician and astronomer Muhammad ibn Musa al-Khwārizmī worked at the House of Wisdom and wrote — the Arabic original does not survive, only later copies and, crucially, Latin translations made centuries afterward — a treatise on arithmetic using the Indian number system. Its content is usually described, following the surviving Latin versions, as a work on Hindu reckoning: how to calculate with the Indian digits and their place-value rules.³ Al-Khwārizmī did not claim the numerals as an Arab invention. He identified them, in the very framing of the work, as belonging to India. That is worth sitting with, because it means the single most consequential transmitter of this number system into the Islamic and then the European world is also the earliest surviving written source to name, explicitly, where it came from.

That naming discipline did not survive the numerals' journey through the Islamic world any better than it survived their journey out of it — the same region that transmitted the digits accurately went on to be governed, converted, and Arabized at three different speeds after the seventh-century conquests, a story with its own stubborn insistence on getting the sequence of events right that is traced separately in The Conquest Took Twenty Years; the Conversion Took Four Centuries. The Arabic text reached medieval Latin Europe only through translation, beginning in the twelfth century, well after al-Khwārizmī's own lifetime, and the surviving Latin manuscripts carry a title that says the same thing his Arabic did: a work customarily rendered as al-Khwārizmī on the numbers — or the art of reckoning — of the Indians. Historians of medieval science generally attribute the translation to the twelfth-century Latin translation movement centered on figures such as Adelard of Bath, though the precise translator and date of any individual surviving copy is genuinely contested among specialists, and that uncertainty is better flagged than assigned with false confidence.⁴ What is not contested is the outcome: Latin Christendom's first extended encounter with this arithmetic came bundled with a title crediting India, transmitted by a scholar working in Arabic.

J. L. Berggren's account of medieval Islamic mathematics is careful on this point in a way worth preserving: the Islamic world's contribution here was not authorship of the numerals but a set of real, substantial achievements built on top of them — systematizing arithmetic algorithms for the Indian digits, extending algebraic method (a word and a discipline that come from al-Khwārizmī's other major treatise, on the science of restoring and balancing quantities), and refining the whole apparatus of calculation that later Latin Europe would inherit as a package.⁵ The numerals rode inside that package. They were the cargo, not the ship.

Al-Khwārizmī did not claim the numerals as an Arab invention; he named them, in the framing of his own treatise, as India's.

The European end of the chain has one especially well-documented link: Leonardo of Pisa, known to history as Fibonacci, the son of a Pisan merchant posted to Bugia, a trading city on the North African coast. Fibonacci later wrote that he learned this arithmetic there as a young man, doing business among Arab merchants and their methods of calculation, and that he found it superior to the Roman numerals and abacus-counting still standard in Italy. In 1202 he published the Liber Abaci, a book that put the Indian digits, place-value arithmetic, and practical commercial calculation in front of an Italian merchant readership for the first time at length, and it is generally credited as the single most influential vehicle for the system's spread through European commerce over the following two centuries.⁶ That the decisive push came from merchant bookkeeping rather than from philosophy or astronomy is not a coincidence of this one case; record-keeping has been the practical engine behind number systems since long before Brahmagupta, a pattern laid out from its earliest evidence in Writing Was Invented by Accountants. Fibonacci learned the system from Arab traders, in an Arab city, doing arithmetic in the Arabic-Islamic style — which is exactly why, when the habit reached the rest of Europe, it took the name of the teachers rather than the source those teachers had themselves named. "Arabic numerals," in thirteenth-century Pisa, was a perfectly accurate description of where the merchant got them. It just was not a description of where they began.

Two alphabets, one ancestor

Here is the question this essay is really built to answer, because it is the one people actually ask when they notice the naming problem: if these are Arabic numerals, why do so many Arabic-language documents use different-looking digits — ٠ ١ ٢ ٣ ٤ ٥ ٦ ٧ ٨ ٩ — instead of the 0–9 on this page? The short answer is that some do and some don't, and the split runs along a fault line inside the medieval Islamic world itself, not along any boundary between "Arab" and "Western" number systems.

The Indian numeral forms did not travel into the Islamic world as a single fixed alphabet. As they spread from the Abbasid heartland outward, two regional glyph traditions diverged from the same Indian originals — Georges Ifrah's sweeping history of numeral systems traces both branches back to shared Indian ancestry, though Ifrah's book is a popular, wide-ranging survey that specialists have criticized on points of detail and precise dating, so it is cited here for the broad shape of the split rather than for any fine-grained claim.⁷ In the Islamic west — the Maghreb and, by extension, Islamic Spain — a set of forms often called ghubār numerals took hold, a word usually translated "dust," possibly referring to a dust-board or sand-tray method of working calculations. It was this western branch, transmitted through Iberia and through traders like Fibonacci's teachers in North Africa, that passed into Latin Europe and, over centuries of scribal drift, settled into the 0–9 shapes standard today.⁸

In the Islamic east — Egypt, the Levant, Iraq, and across to the Gulf — a separate branch of the same Indian-derived forms developed differently and is still in everyday use: ٠١٢٣٤٥٦٧٨٩, sometimes called Eastern Arabic numerals, or in Persian-influenced regions a further variant with different shapes for four, five, and six. Both branches are, structurally, the identical decimal place-value system with the identical zero-as-number logic Brahmagupta set out — same rules, same columns, different handwriting. Menninger's history of number words and number symbols documents this kind of glyph drift as an ordinary, almost inevitable feature of any writing system carried across centuries and regions by hand-copying scribes with no printing press to hold the shapes still.⁹

The geography left a legible scar on the modern map. Morocco, Algeria, and Tunisia — the old Maghreb, the region ghubār numerals spread through and out of — largely use the western 0–9 forms today, the same digits used across Europe and much of the rest of the world, precisely because that is the branch their own region kept using. Egypt and the Arab countries further east largely use the eastern forms. So the honest answer to "why don't Arabs use Arabic numerals" is that a great many of them do, in the specific sense that "Arabic numerals" was always a European name for the western branch — and the branch actually used across most of North Africa today is that same western branch, still legible on the map as the medieval fracture line it always was.

Naming follows the trade route

Step back from numerals entirely and a pattern emerges that is bigger than any one of these stories: things get named for whoever handed them to you, not for whoever made them. It is such a reliable habit that once you start checking, it turns up in places that have nothing to do with mathematics.

The clearest case is a bird. The turkey is native to Mexico and Central America, domesticated by Mesoamerican peoples long before any European ship reached the Americas, and yet almost no language that adopted it as food kept anything resembling that origin in its name. English calls it turkey, after the country, apparently because traders — some via Ottoman territory, some via routes that got the geography confused with African guinea fowl already associated with Turkish merchants — were the last hop before the bird reached English tables. Turkish itself, by contrast, calls the bird hindi, meaning "Indian," blaming India for a bird that has never lived there. French calls it dinde, a contraction of poule d'Inde, "hen from India" — the same misattribution baked directly into the word. Arabic commonly renders it along the lines of dīk rūmī, roughly "Roman," or more precisely Byzantine, fowl, using rūmī in its old sense for the eastern Roman world — yet another wrong direction, and a different wrong direction than French or Turkish chose. Four language communities, four confident answers, four different countries blamed, and not one of them is Mexico. Every language named the bird for the last leg of the trade route it happened to receive it on.¹⁰

Things get named for whoever handed them to you, not for whoever made them — a bird from Mexico blamed on Turkey, India, and Byzantium in three different languages, never on Mexico in any of them.

The pattern holds in metallurgy. What Europeans called Damascus steel — blades with a distinctive rippled, watered surface pattern prized across the medieval Islamic world and, via the Crusades and later trade, in Europe — was not smelted in Damascus. The steel itself, a high-carbon crucible steel known as wootz, was produced in South India and Sri Lanka and exported westward as raw ingots along established trade networks; smiths in Damascus and other Middle Eastern centers imported the ingots and forged them into the blades that carried the city's name. The steel took the name of the market where it was worked and sold, not the region where the metal itself was made — the same displacement, run through iron instead of arithmetic.¹¹

And it holds closer to home than either of those. Roman numerals — I, V, X, L, C, D, M — are called Roman because Rome is where the system reached the form and the dominance that made it the standard reference across the later Latin world. But the Romans did not invent the underlying symbols from nothing; the numeral forms have real antecedents in Etruscan numerals, the counting system of the civilization that preceded and directly influenced early Rome in the same stretch of Italy. "Roman numerals" is a system Rome popularized and stabilized, working from marks with a documented Etruscan ancestry — a smaller-scale, closer-to-home version of exactly the pattern the Indian digits went through on a much larger stage.¹²

This list is kept disciplined rather than padded, because the temptation with a pattern this satisfying is to reach for every half-remembered example and let the essay get looser than its evidence. Several candidates that could not be verified to a firm standard have been left out — plausible etymological folklore is not the same thing as a documented trade history, and this essay only keeps the three above because each has a real paper trail behind it, not just a good story.

What we actually took from him

Which brings the essay back to al-Khwārizmī, and to the one piece of this whole transmission chain where the credit did land on the right name, by accident, through a different kind of drift entirely. Europe did not remember al-Khwārizmī as the source of the digits — those went to "the Arabs" in general, an already-inaccurate collective credit for a system he himself had labeled Indian. But his own name survived anyway, attached not to the numerals but to the method: Latin readers of his arithmetic treatise, and of the tradition of step-by-step calculation it represented, rendered his name as Algorismus, and the word drifted through medieval Latin and into English as algorism and eventually algorithm — a step-by-step procedure for calculation, now the term for the reasoning that runs every computer on earth.

That is the real shape of what happened here, and it is a shape worth stating plainly rather than leaving as a curiosity. The Islamic world's actual, well-documented, substantial contribution to this history was not the invention of the digits — nobody in the chain, including al-Khwārizmī himself, ever claimed it was. The contribution was the systematic method: algorithms for arithmetic, the development of algebra as a discipline, the refinement and transmission of a workable calculating technology across three continents and five centuries, carried by scholars, traders, and translators who kept the machinery running long enough for Fibonacci to meet it in a North African trading city and carry it home. Europe named the digits for the messenger and, centuries later, almost by linguistic accident, gave the messenger's own name to the thing he actually built. The numerals are Indian, and the Arabic-speaking world that transmitted them has been saying so, in its own language, the entire time — al-arqām al-hindiyya, unchanged, while everyone downstream argued about a name that was never really in question at the source.