People by century: organized lists and source notes

Sorting people into centuries looks like the one job on a reference site that cannot go wrong. It goes wrong three times before you have finished the first row: in the arithmetic, in the choice of which date to sort on, and in the calendar the date came from.

What to take away

Professional periodization is usually bounded by events rather than by round numbers, which is why a historian's "eighteenth century" may begin and end years away from 1701 and 1800.
Dates arrive from several systems and cannot be compared until they are converted.
Reconstructed dates enter lists as though they were registered ones.
Count people per century in any large biographical dataset and the counts rise steeply toward the present.

The arithmetic

Because the common era numbering has no year zero, the first century begins with year 1 and ends with year 100. Every century after it is offset the same way: the nineteenth century is 1801 to 1900, the twentieth is 1901 to 2000.

Ordinary usage ignores this and treats "the 1900s" as the twentieth century. Both conventions are in circulation, they differ by exactly one year at each end, and a dataset built from several sources will contain both. Pick one, write it in the documentation, and check the boundary years by hand: they are the only rows where the choice shows.

Across the era boundary, the year before AD 1 is 1 BC. There is no year between them, so any span crossing the boundary is one year shorter than subtraction suggests.

Which date puts a person in a century?

This is the real decision, and it has no correct answer, only a declared one.

Sort on	What the resulting list means	Who it misplaces
Birth year	Cohort: people who entered the world in this span	Anyone whose work happened entirely in the next century
Death year	People whose lives closed in this span	Anyone who did everything decades before dying
Floruit	People documented as active in this span	Anyone whose activity is undated; anyone active across a boundary
Date of the work	The period the output belongs to	The person, who may belong to two or three lists

A person born late in one century and productive across the next appears in a birth-sorted list and a work-sorted list under different headings. Neither list is wrong. A reader who assumes they are the same list will conclude that a source is unreliable.

Long lives make this worse, not better. Someone born in the last years of a century, working through the next, and dying early in the one after touches three centuries and belongs, on the usual conventions, to one.

The practical fix: store a span, not a point. Give each person a start and an end for the basis you are using, and let the list be built by overlap rather than by a single sort key. Then say in plain words what membership means, "documented as active at any point within the span" is a different list from "born within the span," and both are defensible.

Historians rarely use calendar centuries

Professional periodization is usually bounded by events rather than by round numbers, which is why a historian's "eighteenth century" may begin and end years away from 1701 and 1800. The habit of stretching or shortening a century to fit a coherent period is deliberate: the interesting boundaries are wars, revolutions, treaties, epidemics, and technological changes, and those do not fall on the hundreds.

This matters when you import from scholarly sources. A work described as covering a century may cover a period the author has defined, stated once in the introduction, and used consistently thereafter. Copying its contents into a calendar-century bucket silently changes what the list is.

Not every date is in the same calendar

Dates arrive from several systems and cannot be compared until they are converted. Where the question is about origins rather than events, the heritage pillar sets out why conversion is the least of the problems.

The Julian and Gregorian calendars ran in parallel for a long period because jurisdictions adopted the reform at different times. The same day carried different dates depending on where the writer was, which is why careful transcriptions record Old Style and New Style dates together.
Year-start conventions varied, so a date early in the year can belong to either of two years depending on the convention in use.
Lunar and luni-solar calendars have years of different length from the solar year, so a century in one of them is not the same span as a century in another, and their year numbers do not map onto Gregorian ones by addition.
Era systems other than the common era count from different origins, and some restart with a reign or a dynasty.

Conversion tables exist for all of these and are the tool for the job. What you must not do is subtract, add a constant, and move on. Record the source calendar next to the converted value, so a later editor can check the conversion rather than re-deriving it.

Precision the source never claimed

Reconstructed dates enter lists as though they were registered ones. Watch for:

Circa dates, derived from an age at death, an apprenticeship term, a matriculation age, or a stated age in testimony: all of which were routinely rounded.
Ranges collapsed to a midpoint, which is convenient for sorting and invents a year the source did not give.
A baptism date used as a birth date, which is close but not identical, and the gap varied by custom and by circumstance.
A first-mention date treated as a start date, which really marks the first surviving document, not the beginning of an activity.

Keep the qualifier in the data, not just in the prose. A field that stores 1450 for a source that said "c. 1450" has destroyed information that no later reader can recover.

Why every century looks busier than the last

Count people per century in any large biographical dataset and the counts rise steeply toward the present. It is tempting to read something into the shape. There is nothing in it about people, and the same curve appears when the counting is done by country, by state or by city.

The curve is survivorship bias drawn as history. It is produced by record survival, by the growth of printing and then of routine bureaucratic registration, by the expansion of who gets written about at all, and by digitization programs that have concentrated on recent, well-cataloged, machine-readable material. Add the simple fact that reference works are compiled by people who know their own period best.

So a per-century count is a measure of documentation density. It supports statements about the record. It supports no statement about how many notable people lived, which is not a quantity any dataset contains.

The same reasoning kills the tempting follow-up question, which century, region, or population "produced" the most of anything. The answer would be an artifact of archives, and the question itself is one this site does not treat as answerable.

Categories do not travel backwards

A century list invites you to apply one set of labels across the whole span, and most useful labels are younger than the people you are applying them to. Occupations, disciplines, national identities, and institutional categories were coined at particular moments and carried different meanings before and after. Placing a person under a modern heading is an editorial act; it can be a reasonable one, but it should be visible as a choice rather than presented as a property of the person. The profession pages work through this problem in detail.

Building a century list that survives review

Declare the century convention, including the boundary years.
Declare the assignment basis: birth, death, activity, or work.
Store spans with a start and an end, and derive membership from overlap.
Store the source calendar and the original date string beside every converted value.
Preserve approximation markers as data, and never sort on an approximate value as though it were exact.
Record the record type each date came from, since a baptism, a registration, and a scholarly reconstruction are three different kinds of evidence.
State in the page itself that the list is bounded by what survives, not by what happened.
Mark disputed dates as disputed and give both, rather than picking the one that sorts more neatly.

Bottom line

A century is a container the past never used. Putting people into one requires four declared decisions (convention, basis, calendar, and precision), and the value of the list depends entirely on whether those decisions are written down where the reader can see them. When they are not, you have a sorted list whose sort key nobody can reconstruct. The record-verification guide covers where these dates come from in the first place.

Common questions

Which date should place a person in a century?

Whichever your index has declared, applied consistently and stated on the page. Birth date and period of activity give different answers for anyone with a long life.

Why not use named periods instead of calendar centuries?

Named periods carry an argument about what changed and when, and they are defined differently by different fields and regions. Centuries are arbitrary and at least everyone means the same thing by them.

Is it safe to average two approximate dates?

No. The average is a point the sources never asserted, and once printed it gets cited as if it were recorded. Store the range.

Does a rising count per century mean anything at all?

It means more records survive and more have been digitized. Treat the curve as a description of the archive, not of the past.