Roman Numeral Date Converter

What is a Roman numeral date?

It is a date written with the three parts converted to Roman numerals and joined by a separator: 25 December 1994 becomes XXV·XII·MCMXCIV. It is the form used for tattoos, wedding invitations, cornerstones and the copyright line at the end of films, and it is entirely a modern convention.

The converter above takes a day, a month and a year and gives you the numerals in whichever order you want, with whichever separator you want. It also tells you two things nobody else will: whether your date is ambiguous, and how a Roman would actually have written it — which is nothing like the tattoo form.

The numerals themselves come from the same engine as this site's Roman numeral converter, which has been checked exhaustively across all 3,999 representable values rather than by sampling.

How to use it

1. Enter the day, month and year. The year has to be between 1 and 3999, because Roman numerals have no zero, no minus sign, and no standard way of writing 4000 or more. Impossible dates such as 31 February are refused rather than converted into something meaningless.
2. Choose an order and a separator. Day-month-year, month-day-year or year-month-day, joined by an interpunct, a period, a slash, a hyphen or a space. None of these is more correct than the others, which is exactly the problem described below.
3. Read the two notes underneath. The first says whether the date you have chosen can be read two ways. The second gives the authentic Roman form, which is worth a look before you have anything engraved.

132 days a year are ambiguous

Written as XXV·XII·MCMXCIV, nothing in the string says which number is the day and which is the month. For 25 December it does not matter — there is no 25th month, so only one reading is possible. But when both numbers are 12 or under, both readings are valid dates and the numerals cannot tell you which was meant.

That is 12 × 12 combinations, less the 12 where the day and month are the same and the order cannot change anything: 132 days in every year, or about 36% of them. We count this over a real calendar rather than assert it, and get the same figure both ways. IV·VII·MMXXVI is 4 July to a British reader and 7 April to an American one, and both are right.

There is no standard to appeal to here, because the convention is modern and nobody ever wrote one. If the date matters — and on a tattoo it does — the fix is to pick a date where one number is over 12, or to write the month as a word, or simply to tell people the order. The tool flags the ambiguous ones so you find out before the ink does.

Romans did not write dates this way

There was no day-month-year triple in Roman practice. A month had three fixed points — the Kalends on the 1st, the Nones, and the Ides — and every other day was named by counting backwards to whichever came next. The count was inclusive of both ends, so the day before a marker is “two days before”, which Latin gives its own word: pridie.

The Ides of March is 15 March, and everyone knows that one. What is less known is that the Ides is not always the 15th. In March, May, July and October the Nones falls on the 7th and the Ides on the 15th; everywhere else they are the 5th and the 13th. A tempting shortcut is to say the later ones belong to the 31-day months — and that is wrong, because January, August and December also have 31 days and still take the earlier markers. They gained their extra day in the Julian reform without moving their Nones and Ides.

So 25 December 1994 was not XXV·XII·MCMXCIV to anyone in Rome. It was ante diem VIII Kalendas Ianuarias — the eighth day before the Kalends of January, counting inclusively. 31 December is pridie Kalendas Ianuarias. 23 March is ante diem X Kalendas Apriles, because the count runs to April's Kalends and includes both the day itself and the Kalends.

The tool computes this for any date you give it, which is the part of the page you cannot get anywhere else. It also means the modern form is not a translation of anything: it is Roman digits arranged in a modern order, which is a perfectly reasonable thing to want, but not what it is often claimed to be.

Honest limits

The Roman reckoning here uses the Julian calendar's month lengths, which are the ones in the modern calendar too, and applies them to whatever year you type. For a date before the Julian reform of 45 BC, the months were different lengths and the answer would not be right; for a date after 1582 you are looking at what a Roman would have called a day in a calendar Romans never used. It is a translation exercise, and an anachronistic one in both directions.

Latin also declines. The marker days are given in the ablative when the date falls on them (Idibus Martiis) and the accusative after ante diem or pridie (Kalendas Ianuarias), which is why the month name changes shape between answers. That is correct Latin rather than an inconsistency.

The year limit is real, not arbitrary. There is no agreed way to write 4000 or more in standard Roman numerals — the overline convention for multiplying by a thousand exists, but it is not consistently rendered in plain text — so the tool stops at MMMCMXCIX. Years before 1 are not representable at all, since the system has no zero and no sign.

Why is it free?

Because it is arithmetic and a lookup table. Everything runs in your browser, nothing you type is uploaded, and there is no server involved in converting a date.

So there is no account, no sign-up and nothing held back. The engine and its 1,161-assertion test suite are in the repository beside the page. The Roman reckoning is checked against dates whose answers are known independently of the code — the Ides of March, the day before the Kalends of January — because the first summary of the rule we read contained a worked example that was simply wrong, and anchoring on known answers is what catches that.