What a Roman numeral actually encodes
A Roman numeral is an additive tally with a compression trick, not a positional number system. Seven letters carry fixed values — I is 1, V is 5, X is 10, L is 50, C is 100, D is 500 and M is 1,000 — and the value of a numeral is the sum of its letters. MDCLXVI is 1,000 + 500 + 100 + 50 + 10 + 5 + 1 = 1,666, which is the one year that uses each symbol exactly once in descending order.
The trick is the subtractive pair. Writing 4 as IIII is legal and appears on clock faces to this day, but the classical convention writes it as IV: a smaller symbol placed before a larger one is taken away rather than added. Six pairs are sanctioned — IV, IX, XL, XC, CD, CM — and they cap any run of repeated letters at three.
The consequence that matters for a converter is that position carries no weight of its own. In 2,026 the digit 2 means two thousand because of where it sits; in MMXXVI the two Ms mean two thousand because M means a thousand wherever it appears. There is no zero, no fraction and no sign, so the system covers exactly the positive whole numbers. That is why this tool refuses 0 and refuses negatives rather than inventing a symbol for them. If you need a different base rather than a different alphabet, the number base conversion calculator handles binary, octal and hexadecimal, and the number to words converter spells a figure out in English for cheques and contracts.
The subtractive rule, and how the converter applies it
Reading a numeral takes one pass from left to right. Look at each symbol and compare it with the symbol immediately to its right. If the right-hand neighbour is worth more, subtract the current symbol; otherwise add it. The final symbol has no right-hand neighbour, so it is always added. That single comparison is the whole algorithm, and it is the formula shown above: N is the sum of every symbol value sk carrying a sign εk of −1 or +1.
Writing a numeral is the same rule run backwards, and it is easiest as a greedy pass over thirteen values: 1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1, paired with M, CM, D, CD, C, XC, L, XL, X, IX, V, IV, I. Take the largest value that still fits in what remains, write its symbols, subtract it, and repeat. Because the four subtractive pairs sit in the list alongside the plain symbols, the greedy pass produces the classical spelling without any special-casing.
An equivalent way to see it, and the one the breakdown table uses, is digit by digit. Split the number into thousands, hundreds, tens and units, convert each digit on its own, and concatenate. 1,987 is 1,000 + 900 + 80 + 7, which is M + CM + LXXX + VII = MCMLXXXVII. The two methods agree for every value from 1 to 3,999, which is a useful check when you are doing it by hand.
The two rules are not symmetric in strength. Reading accepts far more than writing produces: IIII, VIIII and even IC all read cleanly under the neighbour comparison, yet none of them is what the greedy pass emits. The converter therefore does both — it reads whatever you type, then re-writes the value in standard form and tells you in the Form check tile whether the two agree.
Worked example: 1987 to MCMLXXXVII, and MCMLXXXIV back to 1984
Going forward. Convert 1,987 by splitting it into place-value groups and converting each one.
- Thousands. 1,000 fits once, so write
Mand 987 remains. - Hundreds. 900 is one of the subtractive pairs, so write
CM; 87 remains. - Tens. 80 is 50 + 10 + 10 + 10, so write
LXXX; 7 remains. - Units. 7 is 5 + 1 + 1, so write
VII; nothing remains. - Join them.
M+CM+LXXX+VII=MCMLXXXVII, which is ten letters (M, C, M, L, X, X, X, V, I, I).
Check it by adding the pieces back: 1,000 + 900 + 80 + 7 = 1,987.
Going back. Read MCMLXXXIV one symbol at a time, comparing each with its right-hand neighbour.
M= 1,000, next isC= 100, which is smaller, so add 1,000. Running total 1,000.C= 100, next isM= 1,000, which is larger, so subtract 100. Running total 900.M= 1,000, next isL= 50, so add. Running total 1,900.L= 50, next isX= 10, so add. Running total 1,950.X,X,X= 10 each, each followed by something no larger, so add three times. Running total 1,980.I= 1, next isV= 5, which is larger, so subtract 1. Running total 1,979.V= 5, last symbol, so add. Running total 1,984.
Written as a single sum: 1,000 − 100 + 1,000 + 50 + 10 + 10 + 10 − 1 + 5 = 1,984. Note how close MCMLXXXIV (1984) is to MCMLXXXVII (1987) — numerals for nearby years can differ by several letters, which is why copying a date off a building by eye is unreliable.
The seven symbols, the six subtractive pairs, and the overlined values
| Symbol | Value | Notes |
|---|---|---|
I | 1 | Repeats up to three times |
V | 5 | Never repeated |
X | 10 | Repeats up to three times |
L | 50 | Never repeated |
C | 100 | Repeats up to three times |
D | 500 | Never repeated |
M | 1,000 | Repeats up to three times in standard form |
IV | 4 | Subtractive pair |
IX | 9 | Subtractive pair |
XL | 40 | Subtractive pair |
XC | 90 | Subtractive pair |
CD | 400 | Subtractive pair |
CM | 900 | Subtractive pair |
V̅ | 5,000 | Vinculum form of V |
X̅ | 10,000 | Vinculum form of X |
L̅ | 50,000 | Vinculum form of L |
C̅ | 100,000 | Vinculum form of C |
D̅ | 500,000 | Vinculum form of D |
M̅ | 1,000,000 | Vinculum form of M |
Only I, X and C are ever subtracted, and only from the next two values up — I from V or X, X from L or C, C from D or M. That is why IL for 49 and IC for 99 are not standard. M is never subtracted in the 1–3,999 range because no plain symbol is worth more.
Common numbers and years in Roman numerals
| Number | Numeral | Number | Numeral |
|---|---|---|---|
| 1 | I | 100 | C |
| 4 | IV | 400 | CD |
| 5 | V | 490 | CDXC |
| 9 | IX | 500 | D |
| 14 | XIV | 900 | CM |
| 19 | XIX | 1,000 | M |
| 40 | XL | 1,666 | MDCLXVI |
| 44 | XLIV | 1,984 | MCMLXXXIV |
| 50 | L | 1,987 | MCMLXXXVII |
| 90 | XC | 1,990 | MCMXC |
| 99 | XCIX | 2,026 | MMXXVI |
| 42 | XLII | 3,999 | MMMCMXCIX |
3,999 is the largest number the standard form reaches, because 4,000 would need a fourth M.
How to read the Form check result
Standard form means the numeral you entered is exactly what the greedy pass produces, and the value is between 1 and 3,999. This is the spelling to use for a publication date, a monument, a legal document or a tattoo. It is also the only spelling that round-trips: convert it to a number and back and you get the same string.
Non-standard form means the numeral reads unambiguously but is spelled differently from the modern convention. IIII for 4 is the famous example — it is standard on clock dials, where it balances VIII opposite it. VIIII for 9 appears throughout Roman-era inscriptions. The converter still returns a value for these, and the warning tells you which standard numeral covers the same value. Treat the flag as a note about convention, not as an error.
Outside the standard 1–3,999 range means the value is real but the classical alphabet cannot express it without either repeating M more than three times or reaching for the overline. Choose deliberately here: repeated Ms are readable to anyone, while an overline is compact but relies on the reader knowing the convention and on your typeface rendering the combining mark.
Cannot be written in this notation covers zero, negatives, fractions and anything past the notation's ceiling. The Romans expressed fractions on a separate duodecimal system of dots and used the word nulla where a zero was needed, so there is no numeral to return.
The vinculum and how to type it
The vinculum is a bar drawn over a letter to multiply it by 1,000, so V̅ is 5,000 and X̅ is 10,000. It lets the seven letters reach 3,999,999 — M̅M̅M̅C̅M̅X̅C̅I̅X̅CMXCIX — rather than stopping at 3,999. Unicode encodes the bar as U+0305 combining overline, typed after the letter it covers, and separately provides ready-made numeral characters in the Number Forms block at U+2160 onward. This converter reads either the combining overline or a plain underscore typed before a letter, so _V and V̅ both give 5,000. Set the notation selector to vinculum to get overlined output; leave it on standard and 5,000 comes back as MMMMM.
Mistakes that produce a wrong numeral
- Subtracting from the wrong symbol. Only I, X and C are ever placed in front of a larger letter, and only in front of the next two values up: I before V or X, X before L or C, C before D or M. 49 is
XLIX, notIL; 99 isXCIX, notIC. This converter readsILas 49 and then flags it, because the reading rule and the writing rule are not the same rule. - Repeating a five-symbol. V, L and D never repeat, because
VVis simplyX. Any numeral containingVV,LLorDDcomes back flagged as non-standard. - Reading a date as one number.
XXV·XII·MMXXVIis three numerals side by side, not a single value. Type it with separators and this tool converts each part on its own; run the letters together and you get one very large number instead. - Losing the overline. Copying an overlined numeral into a system that strips combining marks turns 10,000 into 10 silently. If the destination is plain ASCII, write the number out or repeat the M instead.
- Expecting a zero. There is no symbol for zero, so a value such as 1,005 has no placeholder between the M and the V — it is
MV, which looks alarmingly short but is correct.
Where the numerals are still used, and what to use instead
Roman numerals survive in places where a number needs to look formal or needs to be visibly distinct from an ordinary count: monarchs and popes, Super Bowls, book front-matter page numbers, chemical group numbers on older periodic tables, clock and watch dials, film copyright lines, and the cornerstones of public buildings. In every one of those settings the numeral is decorative or ordinal — nobody adds two of them together. That is not an accident of taste. Arithmetic on numerals is genuinely awkward, which is why merchants across medieval Europe switched to Hindu-Arabic digits for calculation while keeping Roman numerals for the formal record.
If your problem is representing a number in a different notation rather than a different alphabet, other tools fit better. Use the number base conversion calculator for binary, octal or hexadecimal, the data storage unit converter when the notation question is really about kibibytes against kilobytes, and the significant figures calculator when the question is how much precision a written figure claims. For angles written in a mixed-radix format that behaves much like a Roman numeral — several units concatenated with no positional weight — see the degrees, minutes and seconds converter. And when you need a figure written out for a cheque or a contract, the number to words converter is the one to reach for.
One historical note worth keeping in mind when you read an old inscription: the strict subtractive convention this calculator treats as standard was not consistently applied in antiquity. Roman-era stonework mixes IIII and IV, sometimes on the same monument, and the tidy six-pair system is a later regularisation. So when the Form check tile says non-standard, it is comparing your numeral against modern convention, not accusing a Roman mason of a mistake.
Key terms
- Subtractive pair
- A smaller symbol placed immediately before a larger one, so its value is taken away rather than added. The six standard pairs are IV, IX, XL, XC, CD and CM.
- Vinculum
- A bar drawn over one or more letters to multiply them by 1,000. Written in Unicode as combining overline U+0305 after each letter it covers.
- Canonical form
- The single spelling that the standard writing rule produces for a value. Several numerals may read as 4, but only IV is canonical.