Numbers to numerals and numerals back to numbers, live in both directions, with strict rules that name the letter at fault, a lenient reading for the forms Rome actually used, dates in every separator style, and the working shown chunk by chunk.
2,024 is written MMXXIV.
Type in either box. The other one keeps up as you go, and the box you are not typing in is tinted to show it is the answer.
1 to 3,999 in ordinary numerals; up to 3,999,999 with the overline. There is no zero and no minus sign.
Lower case is fine. Past 3,999 the overlined part is written in brackets here, since an overline cannot be typed: (V) is five thousand.
Strict enforces the modern standard and names the rule that was broken. IIII, VV, IL and MCMM are all refused, each with the reason and the position of the letter at fault.
2,024 is 2,000 + 20 + 4, written MM + XX + IV.
A numeral is read in chunks, not letter by letter. Each row is one chunk of the standard form and the number it contributes.
Standard form is positional in disguise: each decimal digit is written from its own fixed table, then the four are simply run together.
| Place | Digit | Worth | Written |
|---|---|---|---|
| Thousands | 2 | 2,000 | MM |
| Hundreds | 0 | 0 | nothing |
| Tens | 2 | 20 | XX |
| Units | 4 | 4 | IV |
A digit of nought writes nothing at all, which is exactly why there is no symbol for zero. The gap does the work in a decimal system; here the place simply goes unwritten, which is also why the length of a numeral tells you nothing about the size of the number.
Click any row to load it above. Between them these four columns build every numeral below four thousand.
Where all four subtractive pairs of the low numbers show up: IV, IX, XIV, XIX.
XL and XC are the only two subtractions in this column; the rest simply stack.
CD and CM do the same job an octave up. DCCC is the longest of them, at four letters.
Three Ms is as far as ordinary numerals reach. Past that the overline multiplies by a thousand.
The overline, or vinculum, is a modern tidy-up rather than something Rome used consistently. Roman inscriptions more often wrote large numbers with the apostrophus: CIƆ for a thousand, CCIƆƆ for ten thousand, each extra pair of brackets multiplying by ten.
Not from the initials of Latin number words: that story falls apart on the very first symbol. They are tally marks and Etruscan signs that drifted into the shapes of familiar letters.
A single notch cut into a tally stick. The oldest and plainest mark in the system, and the only one whose shape has never changed.
The deeper cut made at every fifth notch, or the shape of an open hand seen as the V between thumb and fingers. Not the initial of anything: quinque starts with a Q.
Two hands, or two V shapes joined point to point (the cut that closed off every tenth notch). Again not an initial: decem starts with a D, which means five hundred.
From an Etruscan sign written like an arrow or a downward chevron, which flattened through ⊥ into L over several centuries of writing it quickly.
From the Etruscan sign for a hundred, a circle crossed like ⊕. Scribes wore it down to something C-shaped, and only then did centum offer a tidy explanation after the fact.
Half of the old sign for a thousand. When CIƆ was cut down the middle the right-hand half, IƆ, was worth five hundred, and IƆ written in one stroke is D. No Latin word for five hundred begins with D.
The late replacement for CIƆ, the encircled sign the Romans themselves used for a thousand. M only became standard in the Middle Ages, helped along by mille. For most of Roman history a thousand was CIƆ or ↀ.
Almost every clock face with Roman hours prints IIII rather than IV, and has done since long before anybody could ask why. The practical answer is symmetry: IIII balances the VIII opposite it, and the dial then divides into four tidy zones of I-shapes, V-shapes and X-shapes. The workshop answer is casting: a set of IIII, VIII and XII needs exactly twenty I moulds, four V and four X, which one cast plate delivers with nothing left over. The story about Louis XIV demanding it is a later invention; IIII clock faces predate him by centuries.
IV for four is not what Rome mostly did. Inscriptions, wax tablets and the Colosseum's own entrance numbers use IIII freely, and subtractive forms appear sporadically, inconsistently, and sometimes in shapes the modern rules forbid outright: IIX for eight, XIIX for eighteen, IXX for nineteen. The tidy six-pair system taught today was regularised by medieval and Renaissance scribes and only made properly rigid by printers. Strict mode here follows that modern standard; lenient mode follows what people actually carved.
Without a zero there are no place values, and without place values there is no column arithmetic: you cannot line up MCMXCIV under MMXXIV and add. Romans did their sums on a counting board or an abacus and only wrote the result down, which is why Roman arithmetic looks so difficult and was in practice no harder than anyone else's. What the notation genuinely could not do is anything algebraic: no negative numbers, no fractions beyond a system of twelfths, and no way to express a very large number without inventing new marks.
Everything on this page is worked out in your browser from a table of seven symbols and four decimal places. Nothing is uploaded and no request is made. Numbers run from 1 to 3,999 in ordinary numerals and to 3,999,999 with a single overline; strict reading enforces the modern standard and names the rule and the position of any letter that breaks it, while lenient reading accepts the historically attested forms and reports the standard equivalent.
The other end of the same problem: numbers too large to write out, handled by exact digit strings rather than by inventing new symbols.
Work out the date you want to engrave before you convert it: days between two dates, or a date a set number of days away.
For anniversary numerals: find the exact year being marked, then bring it back here to write it out.
A Roman numeral looks like a pile of letters and behaves like one, but it is built the same way a decimal number is. Take the thousands digit, the hundreds, the tens and the units, write each one from its own ten-entry table, and run the four results together: 1994 is 1, 9, 9, 4, which is M, CM, XC, IV, which is MCMXCIV. Reading goes the other way, matching the thousands table, then the hundreds, then the tens, then the units, and a numeral is in standard form exactly when that match consumes all of it. That is why strict validation here can be precise about what went wrong rather than just saying no.
A number on the left, a numeral on the right, and the other one keeps up as you type. The box you are not typing in is tinted and labelled as the answer, so it is never ambiguous which value you supplied and which one the page worked out. Lower case is read as capitals.
Strict enforces the modern standard and refuses anything else, naming the rule that was broken and the position of the letter that broke it. IIII, VV, IL and MCMM each get their own explanation. Lenient accepts what was actually carved and typed, then reports the standard form beside the value.
The panel underneath splits the numeral into chunks with the value each one carries, and shows the four decimal places it was built from. For a birthday or an anniversary, the Dates tab writes all three parts at once in the order and separator you choose, and reads a pasted date back.
IL is refused because I may only be taken from V or X, so 49 is XLIX. IIX is refused because only one symbol is ever subtracted. MCMM is refused because nothing may follow CM except the smaller places beneath it. Each message points at the offending letter by position and shows what the standard form would have been.
IIII is four, XIIX is eighteen, IL is forty-nine, VIV is nine. These are attested forms, not typos: the Colosseum's entrances are numbered with IIII and Roman legions wrote XIIX on tombstones. Lenient reads them all, gives the value, and reports the canonical spelling alongside.
1994 is not seven letters, it is M + CM + XC + IV, and the page shows it that way with the value of each chunk and the sum they make. A second panel breaks the number into thousands, hundreds, tens and units, which is what standard form really is underneath.
Day, month and year converted together, in day-first, month-first or year-first order, with a middle dot, a full stop, a slash, a dash or a space between them. Paste a date in and it reads back the other way, checking that the day exists in that month and that year.
Past MMMCMXCIX an overlined numeral multiplies by a thousand, so 5,000 is V̄ and 1,000,000 is M̄. The page renders the overline properly and offers a bracketed plain-text form, (V), for anywhere you cannot type one. It also says plainly that this is a modern tidy-up rather than settled Roman practice.
One to twenty, the tens, the hundreds, the thousands and the overlined range, plus the seven symbols with where each one really came from. Every row is a button: click it and the converter above fills in.
MMXXIV is 2024. It breaks into MM for two thousand, XX for twenty, and IV for four, and those three chunks simply add: 2000 + 20 + 4. The years either side follow the same shape, because only the units chunk is changing: MMXXIII is 2023, MMXXV is 2025, MMXXVI is 2026. It is worth noticing how short these years are compared with the ones before them: 1888 needs thirteen letters as MDCCCLXXXVIII, while every year from 2000 to 2029 fits in six or fewer, which is a large part of why numerals came back into fashion on watches and film credits at exactly the point the millennium turned.
Split the year into thousands, hundreds, tens and units, write each one from its own table, and run them together. 1987 is 1000 + 900 + 80 + 7, so M + CM + LXXX + VII, giving MCMLXXXVII. 1995 is M + CM + XC + V, so MCMXCV. 2003 is MM + III, so MMIII. The pattern that trips people up is the nine hundreds: every year from 1900 to 1999 begins MCM, not MDCCCC, because nine hundred is written by taking a hundred away from a thousand rather than by stacking four hundreds. Type the year into the box above and the page shows exactly which chunk each part of it became.
Almost every clock face with Roman hours prints IIII, and it is deliberate. The strongest reason is visual balance: IIII sits opposite VIII on the dial, and the two long strokes answer each other, dividing the face into four zones of I-shapes, then V-shapes, then X-shapes. There is a manufacturing reason too: a set of hours from I to XII made with IIII needs exactly twenty I moulds, four V and four X, which is one tidy cast plate with nothing left over, while IV breaks the count. The often-repeated story that Louis XIV insisted on IIII does not survive contact with the evidence: clock faces using IIII predate him by several centuries, including the fourteenth-century Wells Cathedral clock. What is true is that IV was also, in antiquity, an abbreviation of IVPPITER (Jupiter), which may have made it an awkward thing to stamp on a dial.
Old. The additive form was the Roman norm and the subtractive form was the exception, not the other way round. Roman inscriptions, wax tablets, milestones and the numbered entrances of the Colosseum all use IIII freely, and where subtraction does appear it appears inconsistently and sometimes in shapes the modern rules ban outright: IIX for eight, XIIX for eighteen, IXX for nineteen. The tidy six-pair system taught in schools was regularised by medieval and Renaissance scribes and then hardened by printers, who needed one answer rather than several. So IIII is wrong by today's standard and perfectly correct by the standard that actually existed in Rome. Strict mode on this page follows the modern rule; lenient mode follows the older practice and tells you what the modern spelling would be.
There are six legal pairs and no others: IV, IX, XL, XC, CD and CM. Three things follow from that list. Only I, X and C are ever subtracted, never V, L or D. Those already exist as five, fifty and five hundred, so putting one in front of a larger symbol says nothing new. A symbol may only be taken from the next two symbols up, which is why IL and IC are refused: forty-nine is XLIX and ninety-nine is XCIX, each place handled separately. And only one symbol is ever taken away, so IIX is not eight; eight is VIII. Break any of those and strict mode here names which one and points at the letter.
Because the system counts rather than positions. Each symbol carries a fixed value wherever it appears (the C in MCM is a hundred and the C in CM is a hundred), so there are no columns to keep, and therefore no gap that needs marking. A zero is the answer to a question the notation never asks. The cost was real, though: without place values you cannot line two numerals up and add them column by column, so Romans did arithmetic on a counting board or an abacus and wrote down only the result. When the concept did arrive, medieval scholars using Roman numerals for calendar work wrote the Latin word nulla, or its initial N, in tables where a zero was needed. Bede's computus tables from around 725 are the usual example. A symbol never followed.
Not from the initials of Latin number words, which is the story everyone hears and the one that falls apart fastest: unus does not start with I, quinque does not start with V, and decem starts with a D that means five hundred. I, V and X are tally marks: a notch, the deeper cut at every fifth notch or the shape of an open hand, and the crossed cut that closed off every tenth. L, C, D and M come from Etruscan signs that Roman scribes gradually assimilated to whichever letter they most resembled. C descends from a crossed circle for a hundred, D is the right-hand half of the old sign for a thousand, and M replaced that sign, CIƆ, only in the Middle Ages. Centum and mille then supplied a tidy explanation after the fact, which is how the folk etymology got started.
Because 4000 would need four Ms and no symbol exists above M. Three of any symbol is the practical limit, and a fourth is what the subtractive pairs were invented to avoid, so MMM is as far as ordinary numerals reach, and MMMCMXCIX is the largest of them. The Romans themselves did have ways past it: the apostrophus wrote a thousand as CIƆ and ten thousand as CCIƆƆ, each extra pair of brackets multiplying by ten, and a bar over a numeral to multiply by a thousand appears in some inscriptions. Neither was standardised. Modern convention settled on the bar, called a vinculum, which this page supports up to 3,999,999. It is a convention adopted for convenience, not a rule inherited from Rome.
A bar over a numeral multiplies it by a thousand. So V̄ is 5,000, X̄ is 10,000, and M̄ is 1,000,000. Anything above 3,999 is written as an overlined part followed by a plain part: 4,021 is IV̄ then XXI, because four thousand is 4 × 1000 and twenty-one is written normally. That gives a ceiling of 3,999,999, since the overlined group itself cannot exceed MMMCMXCIX. An overline cannot be typed on most keyboards, so plain-text conventions stand in for it; this page uses brackets, writing 5,000 as (V), which has the pleasant accident of matching the Roman (I) for a thousand. Copy the answer and you get the proper overlined characters, or the bracketed form, whichever you ask for.
Convert the day, the month and the year separately, then join them with a separator. 14 March 2024 is XIV, III and MMXXIV, usually written XIV · III · MMXXIV, or XIV.III.MMXXIV, or XIV/III/MMXXIV. The order matters more than the separator does: day-first is standard almost everywhere outside the United States, month-first is standard inside it, and a date like III · IV · MMXXIV is genuinely ambiguous between 3 April and 4 March, which is worth settling before anything gets engraved. None of this is Roman practice, incidentally. Romans dated by naming the consuls of the year and by counting backwards from three fixed points in the month, so 14 March was pridie Idus Martias, the day before the Ides of March. The numeral date is a modern design choice, and a good one, but it is ours rather than theirs.
Explore the growing collection for calculations, documents, writing, and everyday work.