Number systems: how the world writes numbers
Every civilisation that wrote things down worked out how to write quantities, and almost none of them arrived at the same answer. Egyptian scribes drew a picture for each power of ten. Babylonian scribes invented place value two thousand years before anyone else and never quite solved the empty column. Roman numerals lasted a millennium and a half without a zero because they never needed one. This is what each system actually did, and why the one everybody now uses came from somewhere specific.
The distinction everything else depends on
Most arguments about numeral systems are really confusions between four words. Getting them apart makes the rest of this page much shorter.
Number
The abstract quantity itself, independent of how it is written or spoken.
Seven is the same number whether written 7, VII, ז, 七 or 111 in binary. Nothing about the quantity changes.
Properties of a numeral get mistaken for properties of the number. 'Ends in a zero' is a fact about base-10 notation, not about the quantity — 10 ends in 0 in decimal and in 1 in binary.
Numeral
A written symbol or group of symbols standing for a number.
VII, 7 and 七 are three numerals for one number.
Digit
A single symbol from a positional system's fixed set. Base 10 has ten digits, binary has two, hexadecimal sixteen.
42 has two digits, 4 and 2. In binary the same number, 101010, has six.
Only positional systems have digits in this sense. Roman numerals have symbols, and I, V and X are not digits — they carry their value wherever they appear, rather than getting it from a position.
Base (radix)
How many digits a positional system uses, and therefore how much each place is worth relative to the one beside it.
Base 10 uses ten digits and each place is worth ten times the last. Base 2 uses two, and each place is worth twice the last.
Place value
The rule that a digit's contribution depends on where it sits. It is the whole idea; everything else follows from it.
In 505 the two 5s are worth 500 and 5. The symbol is identical; the position does all the work.
Positional notation
A system where position determines magnitude, so a small fixed set of digits writes any number at all.
Decimal, binary, Babylonian sexagesimal and Maya vigesimal are all positional, despite having bases of 10, 2, 60 and 20.
Additive notation
A system where each symbol carries a fixed value wherever it stands, and a numeral's value is the sum of its symbols.
Egyptian numerals: three coils of rope and two strokes is 302, and rearranging them changes nothing.
Additive systems are often called primitive. They are not — they are unambiguous, easy to read and quite unsuited to written calculation, which is a trade-off rather than a failing. Roman accounting worked; it worked on an abacus.
Ciphered notation
Additive, but with a separate symbol for every value in a range — one for 1, another for 10, another for 100 — instead of repeating a symbol.
Ionic Greek and Hebrew both assign letters: κδ is 20 + 4. Compact to write, and it needs twenty-seven symbols where decimal needs ten.
Placeholder
A mark that keeps a place open in a positional system without being a number itself.
The late Babylonian two-wedge sign says 'nothing in this place'. It cannot be added to anything.
A placeholder is not the same achievement as a zero that is a number. Conflating them is what makes 'who invented zero?' unanswerable as usually asked.
Zero as a number
Zero treated as a quantity in its own right, with rules for arithmetic on it.
Brahmagupta's 628 CE rules state what happens when you add, subtract and multiply with zero. That is a different and later step than using a placeholder mark.
Roman numerals
The system nearly everyone half-knows, which is why it is the one where half-knowledge does most damage. Two things worth correcting straight away. Subtractive notation is a convention rather than a Roman rule — IIII and XXXX are common in Roman inscriptions, and the strict six-pair form we teach was regularised in the medieval and early modern periods. And the reason there is no zero is not an oversight: an additive system has no empty column to mark.
Only I, X and C are ever subtracted, and only from the next two symbols up. That restriction is the whole reason 99 is XCIX rather than IC: I may precede V or X and nothing else. Every number profile on this site shows its own Roman form, computed rather than looked up.
The systems, one at a time
Ordered by structure rather than by date: additive systems first, then the ciphered alphabetic ones, then the two positional traditions that developed independently at opposite ends of the world. Each entry says what the symbols were, how they combined, what the system did about zero, and what is genuinely uncertain about it.
Egyptian hieroglyphic numerals
- Script:
- Egyptian hieroglyphs
- Where:
- Egypt and Nubia
- When:
- Attested from around 3000 BCE, near the beginning of Egyptian writing itself, and in use for some three millennia
- Structure:
- Additive
- Base:
- 10
- Place value:
- No
- Zero:
- No zero numeral; a word is attested
On zero. No zero numeral and no need for one. The word nfr marks a zero reference level in architectural drawings and a zero balance in accounts — a word standing in for the idea, not a digit.
Today. None. Read by Egyptologists; written by nobody.
| Value | Symbol | Read as |
|---|---|---|
| 1,000,000 | Heh figureDepicts a kneeling god with raised arms — Heh, associated with unboundedness. | |
| 100,000 | tadpoleDepicts a tadpole or frog. | |
| 10,000 | fingerDepicts a bent finger. | |
| 1,000 | lotusDepicts a lotus plant. | |
| 100 | coil of ropeDepicts a coiled rope. | |
| 10 | hobbleDepicts a cattle hobble — an arch-shaped tether. | |
| 1 | strokeDepicts a single vertical stroke. |
How it composes
- One distinct sign per power of ten, from 1 up to 1,000,000.
- Each sign is repeated as many times as needed and the values are added: 243 is two coils of rope, four hobbles and three strokes.
- There is no place value, so order carries no meaning — scribes grouped repeated signs into neat blocks for legibility.
- Hieroglyphic text runs left to right or right to left (and sometimes in columns); the numerals follow whichever direction the surrounding text takes.
- Hieratic, the cursive script used for everyday documents, compressed these repeated signs into single abbreviated forms — a genuinely different notation built on the same decimal structure.
What is not settled
- Sign shapes vary considerably by period and by scribe; the forms encoded in Unicode follow one standardised palaeographic convention among several.
- The reading of the 1,000,000 sign is debated — it may have meant a specific very large quantity or simply "many", and it is rare in practice.
- Hieratic and demotic numeral forms differ substantially from the hieroglyphic ones shown here and are not modelled.
Display note. Lead with the description — "2 coils of rope, 4 hobbles, 3 strokes" — and treat the hieroglyphs as an enhancement. Most readers have no font covering U+13000–1342F and will see empty boxes. Never make the glyphs the only channel, and give any diagram real alt text describing the signs and counts.
Babylonian sexagesimal numerals
- Script:
- Sumero-Akkadian cuneiform
- Where:
- Mesopotamia — southern Iraq and the wider region
- When:
- Developed from Sumerian precursors around 2000 BCE and used for scribal mathematics and astronomy into the Seleucid period, roughly the last centuries BCE
- Structure:
- Positional
- Base:
- 60
- Place value:
- Yes
- Zero:
- Placeholder only, not a number
On zero. No zero for most of the system's life, and never a zero that was a number. A two-wedge placeholder appears from roughly the sixth century BCE, used between digits; a value could still not end in it unambiguously.
Today. Not written today, but its base survives: 60 minutes in an hour, 60 seconds in a minute, and 360 degrees in a circle all descend from sexagesimal fractions — by way of Hellenistic Greek astronomy, not directly.
| Value | Symbol | Read as |
|---|---|---|
| 1 | vertical wedge (diš)Repeated up to nine times within a place. | |
| 10 | corner wedge (u, Winkelhaken)Repeated up to five times within a place. |
How it composes
- Two wedge signs only: a vertical wedge for 1 and a corner wedge for 10.
- Within one place, the wedges are grouped additively to write any value from 1 to 59.
- Places are positional and read left to right in descending powers of 60: the group on the left is worth sixty times the group on its right.
- Places are separated on the tablet by a space rather than any punctuation.
- There is no sexagesimal point. Whether a numeral meant 2, 120 or 1/30 was settled by context — a working astronomer knew the scale of the quantity being recorded.
What is not settled
- The absence of a place marker in the early system makes some tablets genuinely ambiguous, and modern readings of specific values are sometimes contested.
- How consistently the late placeholder sign was used, and whether any scribe treated it as a number, is debated.
- Why base 60 was chosen is unresolved. Its unusually rich divisibility (2, 3, 4, 5, 6, 10, 12, 15, 20, 30) is the standard explanation; a merger of two earlier counting traditions is another. Neither is established.
Display note. Lead with the wedge description — "2 corner wedges + 4 vertical wedges" — because the cuneiform block (U+12000 onward) renders in very few installed fonts. Show place separation with visible spacing and label which place is which; a reader cannot infer it from the glyphs.
Attic (acrophonic) Greek numerals
- Script:
- Greek capitals, plus compound signs found only as numerals
- Where:
- Attica and other Greek city-states, each with local variants
- When:
- Attested from roughly the 7th century BCE; largely displaced by the Ionic system during the Hellenistic period, though it survives in inscriptions for longer
- Structure:
- Additive
- Base:
- 10
- Place value:
- No
- Zero:
- No zero
On zero. No zero, and no place for one — signs are repeated and summed.
Today. None. Attic numerals are read by specialists, not written by anyone.
| Value | Symbol | Read as |
|---|---|---|
| 1 | IA single stroke rather than a letter name. | |
| 5 | P (pi)First letter of ΠΕΝΤΕ, pente — five. | |
| 10 | D (delta)First letter of ΔΕΚΑ, deka — ten. | |
| 50 | — | pi-delta compoundA small Δ written inside Π: five tens. |
| 100 | H (eta)First letter of ΗΕΚΑΤΟΝ, hekaton — hundred, in the older spelling that still had the rough breathing as a letter. | |
| 500 | — | pi-eta compoundA small Η inside Π: five hundreds. |
| 1,000 | X (chi)First letter of ΧΙΛΙΟΙ, khilioi — thousand. | |
| 10,000 | M (mu)First letter of ΜΥΡΙΟΙ, myrioi — ten thousand, the origin of 'myriad'. |
How it composes
- Each sign is the first letter of the Greek word for its value — hence 'acrophonic'.
- Signs are repeated and added, largest first: ΔΔΠΙΙΙ is 10 + 10 + 5 + 1 + 1 + 1 = 28.
- The five-signs (50, 500, 5,000) are compounds: the sign for 10, 100 or 1,000 written inside the Π for 5.
- There is no subtraction. Four is ΙΙΙΙ.
- Athenian inscriptions also used dedicated signs for currency and weights — drachmas, talents — built on the same acrophonic principle.
What is not settled
- Local variants were the norm rather than the exception — different city-states used different signs, and the 'Attic' set is the Athenian one.
- Attic and Ionic numerals overlapped in use for a long period; the changeover was gradual and uneven, not an event.
Display note. Left to right. The compound signs for 50, 500 and 5,000 live in Unicode's Ancient Greek Numbers block (U+10140 onward) but are blank in most installed fonts, so describe them rather than emitting the codepoints.
Ionic Greek numerals
- Script:
- Greek alphabet, including three letters retained only as numerals
- Where:
- Ionia and Miletus first; then the Greek-speaking world and the Byzantine Empire
- When:
- In use from roughly the 5th–4th century BCE, displacing Attic numerals over the following centuries; still used for ordinal numbering in Greek today
- Structure:
- Ciphered additive
- Base:
- 10
- Place value:
- No
- Zero:
- No zero
On zero. No zero. A ciphered-additive system has nothing for zero to do: each letter already states its own magnitude, so there is no column that could be empty.
Today. Greek still uses these letters as ordinals — chapter numbers, monarch and patriarch regnal numbers, list items — much as English uses Roman numerals.
| Value | Symbol | Read as |
|---|---|---|
| 1 | alpha | |
| 2 | beta | |
| 3 | gamma | |
| 4 | delta | |
| 5 | epsilon | |
| 6 | stigma | |
| 7 | zeta | |
| 8 | eta | |
| 9 | theta | |
| 10 | iota | |
| 20 | kappa | |
| 30 | lambda | |
| 40 | mu | |
| 50 | nu | |
| 60 | xi | |
| 70 | omicron | |
| 80 | pi | |
| 90 | koppa | |
| 100 | rho | |
| 200 | sigma | |
| 300 | tau | |
| 400 | upsilon | |
| 500 | phi | |
| 600 | chi | |
| 700 | psi | |
| 800 | omega | |
| 900 | sampi |
How it composes
- Twenty-seven letters cover 1–9, 10–90 and 100–900, one symbol per value.
- Numbers are written in descending order and added: κδʹ is kappa (20) + delta (4) = 24.
- A keraia (ʹ), a mark like a prime, follows the numeral to show the letters are a number and not a word.
- Thousands reuse the unit letters with a mark below and to the left: ͵α is 1,000.
- Because each letter carries its own value, order is convention rather than meaning — δκ would still be 24, but nobody wrote it that way.
What is not settled
- The sign for 6 appears variously as digamma (ϝ), stigma (ϛ) or the two-letter στ, differing by period and by scribe. Numberwise shows stigma, the commonest printed form.
- Koppa for 90 has several shapes across periods (ϙ, ϟ), none of them wrong.
- Whether the Ionic system was adapted from a Phoenician-influenced model or devised independently in Miletus is not settled.
Roman numerals
- Script:
- Latin alphabet
- Where:
- Rome and its empire; then Europe, and now internationally as a stylistic convention
- When:
- c. 5th century BCE onwards; displaced for calculation by Hindu-Arabic numerals in Europe between roughly the 13th and 16th centuries
- Structure:
- Additive, with subtractive pairs
- Base:
- 10
- Place value:
- No
- Zero:
- No zero numeral; a word is attested
On zero. No zero symbol, and no need for one: a non-positional system has no empty column to mark. The Latin word nulla (abbreviated N) appears in medieval Easter tables where a zero value was required.
Today. Still in everyday use for ordinal and decorative purposes — clock faces, book preliminaries, monarch and pope regnal numbers, film copyright dates, Super Bowl editions, and outline numbering. Nobody calculates with them.
| Value | Symbol | Read as |
|---|---|---|
| 1 | IProbably a tally stroke. | |
| 5 | VOften explained as an open hand — the five-count. | |
| 10 | XCommonly read as two Vs, one inverted; the derivation is not settled. | |
| 50 | L | |
| 100 | CReinforced by centum, Latin for a hundred. | |
| 500 | D | |
| 1,000 | MReinforced by mille, Latin for a thousand. |
How it composes
- Symbols are written in descending order and added: XVII is 10 + 5 + 1 + 1 = 17.
- A symbol may repeat up to three times in the canonical modern form: XXX is 30, and 40 is written XL rather than XXXX.
- Only I, X and C are subtracted, and only from the next two symbols up — giving exactly six subtractive pairs: IV, IX, XL, XC, CD, CM.
- That restriction is why 99 is XCIX (90 + 9) and not IC: I may only precede V or X.
- V, L and D are never repeated and never subtracted; each already represents a half-step, so DD would mean the same as M.
What is not settled
- Subtractive notation was never applied consistently in antiquity. IIII and XXXX are common in Roman inscriptions, and the strict six-pair form taught today was regularised in the medieval and early modern periods.
- The origin of the symbols is contested. The tally-mark account (I, V as the notch cut for five, X as a crossed notch) is well argued but not proven, and the derivation-from-Greek-letters account has its own supporters.
- The vinculum — an overline multiplying a numeral by a thousand — is attested but was never standardised, and competing conventions exist for millions. Numberwise does not generate it.
Hebrew numerals
- Script:
- Hebrew alphabet (right to left)
- Where:
- Wherever Hebrew is written; today chiefly Israel and Jewish communities worldwide
- When:
- Attested from roughly the Hellenistic period, likely on the Ionic Greek model; continuously in use since
- Structure:
- Ciphered additive
- Base:
- 10
- Place value:
- No
- Zero:
- No zero
On zero. No zero. Nothing in the system needs one, and modern Hebrew writes 0 with the Hindu-Arabic digit.
Today. Ordinary numbers in modern Hebrew are written with Hindu-Arabic digits. The letter numerals remain standard for Hebrew calendar dates, Torah and Talmud references, and ordinal numbering — the way Roman numerals survive in English.
| Value | Symbol | Read as |
|---|---|---|
| 1 | alef | |
| 2 | bet | |
| 3 | gimel | |
| 4 | dalet | |
| 5 | he | |
| 6 | vav | |
| 7 | zayin | |
| 8 | het | |
| 9 | tet | |
| 10 | yod | |
| 20 | kaf | |
| 30 | lamed | |
| 40 | mem | |
| 50 | nun | |
| 60 | samekh | |
| 70 | ayin | |
| 80 | pe | |
| 90 | tsadi | |
| 100 | qof | |
| 200 | resh | |
| 300 | shin | |
| 400 | tav |
How it composes
- Twenty-two letters cover 1–9, 10–90 and 100–400.
- Letters are written in descending order and added: כד is kaf (20) + dalet (4) = 24.
- Above 400, tav (400) is repeated: 500 is תק, 800 is תת.
- 15 and 16 break the pattern. Written by the ordinary rule they would be יה and יו, which spell forms of the divine name, so tradition writes טו (9+6) and טז (9+7) instead.
- A geresh (׳) marks a single-letter numeral; a gershayim (״) sits before the last letter of a longer one, signalling that the letters are a number rather than a word.
- Thousands are usually written separately, often with their own geresh, rather than by extending the letter values.
What is not settled
- The 15/16 substitution is universal in practice but its earliest attestation is not precisely dated.
- Final letter forms (ך ם ן ף ץ) are sometimes assigned 500–900 in gematria; this is a gematria convention rather than standard numeral notation, and Numberwise does not use it for writing numbers.
- Whether the system was borrowed from Ionic Greek or developed in parallel is argued both ways.
Chinese numerals
- Script:
- Chinese characters (Han)
- Where:
- China, Taiwan, Hong Kong, Macau and Singapore; the same characters are read with different pronunciations in Japanese, Korean and Vietnamese
- When:
- Number characters appear on Shang oracle bones from around the 13th century BCE and have been in continuous use since
- Structure:
- Multiplicative-additive
- Base:
- 10
- Place value:
- No
- Zero:
- Has a zero, used as a number
On zero. 〇 (and the fuller 零) is a genuine zero, used both as a digit and as a placeholder inside larger numbers. It entered Chinese numeral writing comparatively late — counting rods marked an empty place with a blank space first.
Today. Fully current. Hindu-Arabic digits are used for most arithmetic and prices, while the characters remain standard in prose, dates, formal documents and vertical text. The financial forms are legally required on Chinese cheques and many contracts.
| Value | Symbol | Read as |
|---|---|---|
| 0 | líng | |
| 1 | yī | |
| 2 | èr | |
| 3 | sān | |
| 4 | sì | |
| 5 | wǔ | |
| 6 | liù | |
| 7 | qī | |
| 8 | bā | |
| 9 | jiǔ | |
| 10 | shíTen, used as a named power. | |
| 100 | bǎiHundred. | |
| 1,000 | qiānThousand. | |
| 10,000 | wànTen thousand — the grouping unit, where English groups by thousands. Simplified 万. |
How it composes
- Digits 一 to 九 combine with the named powers 十, 百, 千, 萬.
- A digit written before a power multiplies it, and the terms are added: 四十五 is four tens plus five.
- Ten to nineteen are conventionally written without the leading 一 — 十一 rather than 一十一 — when standing alone.
- Large numbers group by ten thousands (萬), not by thousands, so 100,000 is 十萬, ten wàn.
- The financial forms 壹貳參肆伍陸柒捌玖拾 replace the ordinary digits on cheques and contracts, because the simple characters can be altered with a single added stroke.
- 〇 or 零 marks a skipped power inside a larger number: 一百零五 is 105.
What is not settled
- Whether 〇 was borrowed from Indian notation, developed from the blank space left by counting rods, or arrived by both routes is not settled.
- Regional and register variation is real: 兩 rather than 二 before some measure words, and differing conventions for reading years digit by digit.
- Financial character sets differ slightly between mainland simplified usage and traditional usage in Taiwan and Hong Kong.
Chinese rod numerals
- Script:
- Arrangements of physical counting rods, and their written imitation
- Where:
- China, and adopted in Japan, Korea and Vietnam
- When:
- In use from at least the Warring States period (roughly the 5th–3rd century BCE) until the abacus displaced them, largely by the Ming dynasty
- Structure:
- Positional
- Base:
- 10
- Place value:
- Yes
- Zero:
- Placeholder only, not a number
On zero. An empty place was originally left blank on the counting board. A written 〇 for the empty place appears later, once rod arrangements began to be recorded on paper rather than laid out physically.
Today. None, beyond the study of historical mathematics.
| Value | Read as |
|---|---|
| 1 | vertical rods (units, hundreds, …)One to five rods laid side by side; six to nine add a horizontal rod above worth five. |
| 10 | horizontal rods (tens, thousands, …)The same forms rotated ninety degrees, so adjacent places cannot be confused. |
How it composes
- A fully decimal positional system, worked on a ruled counting board.
- Each place holds one to nine rods; a rod crossing the others counts as five, so 7 is one crossing rod plus two.
- Alternate places alternate orientation — vertical for units, horizontal for tens, vertical for hundreds — which is what keeps neighbouring places legible.
- An empty place is simply left blank on the board.
- Red and black rods (later, differently marked ones) distinguished positive from negative quantities — a working notation for negative numbers well before Europe accepted them.
What is not settled
- Dating the system's origin depends on textual references that predate surviving physical evidence.
- How early the blank place was consistently treated as a zero, rather than a gap needing interpretation, is debated.
Display note. Unicode encodes rod numerals at U+1D360 onward, with almost no font coverage, and inline text cannot convey the alternating orientation that makes the system readable. Present as a diagram with descriptive alt text.
Maya numerals
- Script:
- Maya hieroglyphic writing
- Where:
- The Maya region — southern Mexico, Guatemala, Belize, and western Honduras and El Salvador
- When:
- Attested in the Maya Long Count from the first centuries BCE and used through the Classic period (roughly 250–900 CE) and beyond; the underlying vigesimal counting is shared across Mesoamerica and older than the Maya evidence for it
- Structure:
- Positional
- Base:
- 20
- Place value:
- Yes
- Zero:
- Has a zero, used as a number
On zero. A genuine positional zero, written as a shell, developed independently in Mesoamerica. Whether Maya mathematicians also treated zero as a number to calculate with — as Brahmagupta later did in India — is not clear from the surviving record, which is overwhelmingly calendrical.
Today. Not used for ordinary arithmetic today, though the numerals appear in Maya cultural and educational contexts and several Mayan languages still count vigesimally.
| Value | Symbol | Read as |
|---|---|---|
| 1 | dotUp to four in a place. | |
| 5 | barUp to three in a place. | |
| 0 | shellA stylised shell, marking an empty place. |
How it composes
- Each place holds a value from 0 to 19, written as up to three bars (five each) with up to four dots (one each) above them.
- Places stack vertically, smallest at the bottom; each place up is worth twenty times the one below.
- A shell fills a place that would otherwise be empty — the system's zero.
- Ordinary vigesimal places run 1, 20, 400, 8,000.
- The Long Count calendar uses the same marks with a modified third place worth 18 × 20 = 360 rather than 400, so that one cycle approximates a solar year. Calendar notation and vigesimal arithmetic are not the same system.
What is not settled
- Almost all surviving Maya numerals are calendrical. How the system was used for everyday commercial arithmetic is inferred rather than directly attested.
- Whether the positional zero originated with the Maya or with an earlier Mesoamerican tradition — Olmec or Epi-Olmec — is unresolved; the earliest Long Count dates are not Maya.
- The bar-and-dot numerals also appear in head-variant and full-figure forms, where each number is written as a deity's portrait. Those are the same values in an entirely different visual register.
Display note. Describe the marks — "2 bars + 3 dots over 1 dot" — as the primary channel; Unicode's Mayan Numerals block has almost no font coverage. Vertical stacking carries the place value, so any diagram must show it and any alt text must state which group is the upper place.
Who invented zero?
The question hides two different inventions, and most answers give you one of them as if it were both.
A placeholder
A mark meaning "nothing in this column". Babylonian scribes used one from around the sixth century BCE. It keeps a numeral readable and cannot be added to anything.
A positional zero
A digit occupying a place in its own right. The Maya shell does this, developed with no contact with the Old World — proof the idea was reached more than once.
Zero as a number
Something you can calculate with. Brahmagupta stated the rules in 628 CE. His rule for division by zero does not work — which tells you the question was already being asked.
The full account, with the history of the digit shapes and the arithmetic, is on the profile for 0.
Where the numbers you are reading came from
The name "Arabic numerals" records the route rather than the origin. The numerals and the positional system are Indian; Arabic-speaking mathematicians transmitted and extended them, and Arabic itself calls them al-arqam al-hindiyyah, Indian numerals. Dates below are given as scholarship gives them, which is often loosely.
c. 3rd century BCE · India
Brahmi numerals appear in inscriptions. They have distinct signs for 1–9 but also separate signs for tens and hundreds, so they are ciphered rather than positional — no place value and no zero. The shapes of the modern digits descend from these.
c. 2nd millennium BCE onwards · Mesopotamia
Babylonian sexagesimal notation is already fully positional, and by the later period uses a placeholder sign for an empty place. This is positional notation long before India, but with a placeholder rather than a number.
By the 1st centuries BCE–CE · Mesoamerica
Maya Long Count notation uses a shell as a true positional zero, developed with no contact with the Old World — evidence that the idea was reached more than once.
By the mid-1st millennium CE · India
Decimal place-value notation is in use. Aryabhata's Aryabhatiya (499 CE) works with place value, and the Bakhshali manuscript contains a dot used as a placeholder.
Contested: The Bakhshali manuscript's date is actively disputed. A 2017 radiocarbon study proposed 3rd–4th century CE for some folios; specialists have argued on textual and palaeographic grounds for a considerably later date, and the manuscript is composite, so a single date may not be meaningful.
628 CE · India
Brahmagupta's Brahmasphutasiddhanta gives the first known systematic rules for arithmetic with zero and with negative numbers, treating zero as a number rather than a gap. His rule for division by zero does not work, which is itself informative: the question was being asked.
876 CE · Gwalior, India
An inscription at the Chaturbhuj temple records the numbers 270 and 50 with a small circle for zero — the earliest undisputed dated Indian inscriptional use of the zero digit.
c. 825 CE · Baghdad
Al-Khwarizmi writes On the Calculation with Hindu Numerals, explaining the Indian system to an Arabic-reading audience. His name gives us 'algorithm'; the title of another of his works gives us 'algebra'. Al-Kindi's near-contemporary treatise spreads it further.
10th–12th centuries · Al-Andalus, North Africa and Europe
The numerals reach Europe through Islamic Spain. Gerbert of Aurillac, later Pope Sylvester II, encounters them in the 10th century; Latin translations of al-Khwarizmi follow in the 12th.
1202 · Pisa
Fibonacci's Liber Abaci argues the case for the numerals to European merchants on practical grounds — they make bookkeeping arithmetic tractable in a way Roman numerals never did.
13th–16th centuries · Europe
Adoption is slow and contested. Several Italian cities restricted Hindu-Arabic numerals in official ledgers, partly because they were easier to alter fraudulently than Roman ones — a 0 becomes a 6 or a 9 with one stroke. Printing eventually settled the matter.
Today · Global
Two descendant digit sets remain in wide use: the Western Arabic digits 0123456789, and the Eastern Arabic digits ٠١٢٣٤٥٦٧٨٩ used across much of the Middle East and South Asia. Both come from the same Indian source, and both are positional decimal.
Why bases differ
A base is how many digits a positional system uses, and therefore how much each column is worth. None of the bases in common use was chosen for a mathematical reason.
Base 2 — binary
A switch is reliably on or off and unreliably anything between. Two states is what electronics can hold without error, so the base is a physical constraint rather than a mathematical preference.
Where you meet it: Every digital device; file sizes; the powers of two that recur through computing.
Base 8 — octal
One octal digit is exactly three bits, which suited machines with word sizes divisible by three.
Where you meet it: Unix file permissions — the 755 in chmod 755 is three octal digits, one per permission group.
Base 10 — decimal
Ten fingers. There is no mathematical case for it: 12 divides more evenly and 2 suits machines better. Anatomy settled the question long before anyone thought to compare.
Where you meet it: Ordinary written numbers everywhere; the metric system; percentages.
Base 12 — duodecimal
12 divides by 2, 3, 4 and 6 where 10 divides only by 2 and 5, which matters enormously when goods are split by hand rather than by calculator. One account of its origin counts the three finger joints on each of four fingers with the thumb, giving twelve per hand.
Where you meet it: Dozens; inches in a foot; months; hours on a clock face; old British currency.
Base 16 — hexadecimal
One hex digit is exactly four bits, so a byte is always two characters and the mapping to binary needs no arithmetic. Decimal has no such alignment, which is why nobody writes byte values in decimal.
Where you meet it: Memory addresses; colour codes such as #FF0000; MAC addresses; hashes.
Base 20 — vigesimal
Fingers and toes. Counting the whole body rather than the hands gives twenty, and the pattern appears independently in Mesoamerica, in Celtic languages and in West Africa.
Where you meet it: Maya numerals; French quatre-vingts for eighty; the English 'score'; Danish number words.
Base 60 — sexagesimal
60 divides by 2, 3, 4, 5, 6, 10, 12, 15, 20 and 30, so most fractions of it come out whole. Whether that is why the Babylonians chose it is not established, but it is why it survived.
Where you meet it: Minutes and seconds, of time and of angle; the 360-degree circle.
How these are shown here
Some of these scripts render in any browser and some do not. Egyptian hieroglyphs, cuneiform, Maya numerals and the Attic compound signs are all encoded in Unicode and missing from almost every installed font, so a page that shows only the glyphs shows most readers a row of empty boxes. Numberwise leads with a description in those cases — "2 corner wedges + 4 vertical wedges" — and treats the codepoints as an enhancement for readers whose systems can display them.
Where a system's notation genuinely varies, no single form is presented as the correct one. Attic numerals stop at 49 here because the compound signs above that cannot be shown honestly; Babylonian stops at two places because a third compounds an ambiguity the system already had; Maya stops at 400 because that is exactly where ordinary vigesimal arithmetic and the Long Count part company.
Common questions
Why don't Roman numerals have a zero?
Because they never need one. Zero's job in a positional system is to hold an empty column open — to keep 105 from collapsing into 15. Roman numerals are additive: every symbol carries its own value wherever it stands, so there are no columns and nothing to hold open. Medieval Latin computists writing Easter tables did use the word nulla, sometimes abbreviated N, where a zero was required, but that is a word in a table rather than a numeral.
Why is 4 written IV rather than IIII?
Subtractive notation: a smaller symbol before a larger one is subtracted, so IV is five minus one. It is worth knowing that this is a convention rather than a Roman rule — IIII appears throughout Roman inscriptions, and it survives on most clock faces, where it balances the VIII opposite. The strict six-pair form taught today (IV, IX, XL, XC, CD, CM) was regularised long after Rome.
Who invented zero?
The question has no single answer because it hides two different inventions. A positional placeholder — a mark meaning 'nothing in this column' — was used in Babylonian notation from around the sixth century BCE, and a true positional zero, the shell, developed independently in Mesoamerica. Zero as a number in its own right, with stated rules for arithmetic on it, is the specifically Indian contribution: Brahmagupta set out those rules in 628 CE. All three are real, and they are not the same achievement.
Where did the numbers we use today come from?
India. The digit shapes descend from Brahmi numerals of around the 3rd century BCE, and decimal place-value notation with a zero digit was in use in India by the mid-first millennium CE. Arabic-speaking mathematicians — al-Khwarizmi above all — transmitted and extended the system, which is why Europe called the numerals Arabic; Arabic itself calls them al-arqam al-hindiyyah, Indian numerals. They reached Europe between the 10th and 13th centuries and displaced Roman numerals for calculation over the following three hundred years.
Why does time use base 60?
It descends from Babylonian sexagesimal arithmetic, but not directly. Babylonian astronomers wrote fractions in base 60; Hellenistic Greek astronomers adopted that convention for astronomical calculation; it passed through Islamic and then European astronomy into ordinary timekeeping and angle measurement. Sixty survived because it divides evenly by 2, 3, 4, 5, 6, 10, 12, 15, 20 and 30, so most fractions of an hour come out whole. Why the Babylonians chose it in the first place is not established.
Why do computers use binary?
Because a switch is reliably on or off and unreliably anything in between. Two states is what the physical hardware can hold without error, so base 2 is a constraint of electronics rather than a mathematical preference. Hexadecimal exists on top of it for human convenience: one hex digit is exactly four bits, so a byte is always two characters.
What is the difference between a number and a numeral?
A number is the abstract quantity; a numeral is a symbol standing for it. Seven is the same number whether written 7, VII, ז, 七 or 111. This matters more than it sounds: 'ends in a zero' is a fact about base-10 notation, not about the quantity — 10 ends in 0 in decimal and in 1 in binary.