Features
Part of How to walk a historical figures citation chain back to its source
Where the historical figures data comes from, and its limits
Historical figures data starts with dates: calendar reform, year-start conventions, regnal dating, and a field design that keeps the source readable.
A date looks like the least controversial field in any record about a person. It is often the most: the same day can carry two different year numbers, the same month can be numbered differently, and a printed date can be a conversion someone else performed without saying so.
This page is about storing dates from historical records so that later readers can tell what the document said from what you worked out.
What to take away
- Store the date as written, then store your conversion beside it, never on top of it.
- The year did not always begin on the first of January, so a date in the early months can belong to two years.
- Date precision is data: "1841" and "about 1841" are different facts and need different fields.
- A converted date without a stated conversion rule cannot be checked by anyone, including you.
Four ways a historical date goes wrong
| Problem | What it looks like | What it does |
|---|---|---|
| Calendar reform | Two systems in parallel for a long stretch | Shifts a date by a number of days that varies by period |
| Year-start convention | A year beginning in March, not January | Moves a January to March date into the wrong year |
| Regnal or era dating | "In the twelfth year of" | Needs a reign table, and reigns overlap in disputed periods |
| Silent conversion | A tidy modern date in a printed edition | Hides an editorial decision inside what looks like a source |
Calendar reform is the best known and the least dangerous, because everyone expects it. Documents from the transition period often carry both year numbers for exactly that reason, which is why Old Style and New Style dates appear together in careful transcriptions.
The year-start convention causes more damage, because it produces an error of a whole year in a date that looks perfectly normal. Reforms and conventions were adopted at different times in different jurisdictions, so two documents from neighboring territories in the same month can be dated differently and both be correct.
A field design that survives
- Date as written, verbatim, including the original's own qualifiers.
- Calendar or era stated, if the document states or implies one.
- Converted date, with the rule used, stored separately.
- Precision, as an explicit value: exact day, month, year, decade, or range.
- Basis, meaning what the date is a date of: the event, the registration, or the copy.
The fifth is the one most often skipped and the one that most often produces an argument. A record's date is usually the date of recording, not the date of the event, and for a late registration those can be decades apart.
Precision is not a rounding problem
Padding an approximate date to a specific one produces a dataset that charts smoothly and is partly invented. A decade-level source becomes a mid-decade year, that year gets counted in an annual series, and the series then shows structure that came from the padding rule.
Store the range. Where a chart needs a point, derive the point at chart time from the stored range, and say in the caption how it was derived. The stored value stays honest and the derived one stays reproducible.
Where a project needs to compare dates across systems at all, a single stated internal convention such as the proleptic Gregorian calendar makes the comparison possible, provided the convention is written down and applied only in the derived column.
Dating from other kinds of evidence
Sometimes there is no date, only a bracket. A person appears in one record and is absent from a later one; a document is signed during a known tenure; a place name is used that was only current for a span of years.
These are legitimate and they are ranges, so record them as ranges with the reasoning. National agencies' guidance for reconstructing timelines from record series, for example their notes on military records, shows how much a series can bracket even when it dates nothing directly.
What a date does not establish
A date on a record establishes when something was written down. It does not, by itself, establish where a person was, who they were related to, or what they were. It is a hinge, not a conclusion.
One specific point: no date on this site is used to state anything about a living person's origin or background. That is covered on the heritage pillar.
For how these dates then get bucketed into periods, see the century pillar. For the search methods that produce them, see the historical figures pillar; and for their interaction with jurisdictions, see the country pillar.
Common questions
Should I convert dates at all?
Yes, in a separate column, for sorting and comparison. Never in the column that holds what the document said.
A published edition gives a clean date. Can I use it?
Cite it as the edition's date, and note that editions often convert silently. If the argument depends on the date, go to the original.
How do I record a date that two sources give differently?
Both, each with its source, plus a note on which you prefer and why. A single reconciled date hides the fact that anyone disagreed.
Is a range too vague to publish?
No. A range with a stated basis is more informative than a false point, and readers handle ranges perfectly well when they are labeled as ranges.
