ON THIS DAY

Editors do, working from whatever archive their organisation happens to hold. Every "on this day in history" list is a compiled artefact, and the shape of the collection behind it is visible in the result once you know to look.

That is not a complaint about the format. It is the most interesting thing about it.

Key Takeaways

• The day-by-day chronology began in newspaper almanacs. The World Almanac for 1876 printed a dated "Record of Events" that reads like one New York newspaper's year sorted by date.
• Chase's Calendar of Events started in a Michigan basement in 1957: thirty-two pages, 364 entries, priced at one dollar.
• Wikipedia publishes its rules. Roughly five events run per day, and its guideline says outright that what gets posted is "NOT solely based on what is 'the most important or significant'".
• The BBC's own archive says it shows stories "broadcast by BBC News since 1950". The window is set by what BBC News broadcast, not by the importance of what happened.
• Britain's Calendar (New Style) Act 1750 deleted eleven dates: 2 September 1752 was followed by 14 September, and New Year's Day moved from 25 March to 1 January.

Where does the "on this day in history" format come from?

From almanacs, which were reference books long before they were entertainment.

Most of The World Almanac for 1876 is what you would expect: sunrise and sunset tables, the four seasons timed to the minute, postal rates, the vote in every state. A working reference sold by a newspaper to readers who needed numbers.

At page nine sits "Record of Events", a chronology of the previous year taken date by date. It reads exactly like a modern day-in-history feed:

Jan. 1. Samuel J. Tilden was inaugurated Governor of New-York.

March 12. Moody and Sankey began revival meetings in London.

March 20. In the University boat-race, Oxford beat Cambridge by ten lengths.

The catalogue record gives the publisher as Press Publishing of New York, the house behind the New York World. Read a month of the chronology and that origin is hard to miss, though the inference is mine rather than the book's: Congressional votes come with the tally, Louisiana and New York politics dominate, and European items are sparse. It reads like one newspaper's year sorted by date, and the sorting is the only thing that makes it feel neutral.

The second ancestor is more deliberate. In 1957 William Chase, a librarian in the newsroom of the Flint Journal in Michigan whose job was clipping and filing stories, asked his brother Harrison, a professor in Florida, to help him build a single book of dates. The first edition ran to thirty-two pages, carried 364 entries, cost a dollar and was printed in 2,000 copies from Chase's basement. They advertised it for eleven dollars in the Saturday Review of Literature and did not sell the print run.

Chase's Calendar of Events is still published every year, and by the time of that profile its editions ran to 752 pages and more than 12,000 entries. The file swelled because organisations began asking to be in it, which is a second kind of selection.

Compilation	Who chose the entries	Drawn from	Coverage
The World Almanac for 1876	Press Publishing of New York	The previous year, mostly US politics	One year, reset annually
Chase's Calendar of Events, 1957 onward	Two brothers, then in-house editors	Clippings, reference works and submissions from organisations	The whole calendar year, reissued annually
BBC ON THIS DAY	BBC News editors	The BBC's own broadcast archive	1950 to 2005, plus 1939 to 1945
Wikipedia selected anniversaries	Volunteer editors under a published guideline	Existing Wikipedia articles that meet a quality bar	Any date, refreshed every year

Who decides what appears on the list today?

For one of the most widely reproduced versions, the answer is a body of volunteers following a written rule book, and that rule book is unusually candid.

Wikipedia's selected anniversaries guideline opens by saying that what gets posted is "NOT solely based on what is 'the most important or significant'. Article quality and the mix of topics listed are also considered." Generally five events run per day. Preference goes to round-number anniversaries, and the bar is higher for older material: an event at least a century old is favoured when the anniversary is a multiple of twenty-five, while a younger event qualifies at a multiple of ten.

Two of its criteria decide more than any judgement about significance. The article behind an entry must not be a stub and must be reasonably complete and free of maintenance tags. And the date itself "must be cited to a reliable source with an inline citation".

Read that as a filter and it is a strict one. An event with no decent encyclopedia article cannot appear, however important it was. An event whose exact date nobody has cited cannot appear either. The guideline even asks editors to spread entries across "geographical areas (e.g., more than just the Anglosphere)", which is an admission that the raw pool leans one way and has to be corrected by hand.

The BBC's ON THIS DAY site draws its boundary somewhere else and is equally open about it. Its About page calls the site "a showcase of some of the most significant as well as some of the quirkier stories broadcast by BBC News since 1950", offering reports "drawn from the years 1950-2005 and 1939-45", written "as if the event had only just occurred".

Every clause there is an edge. The main run covers 1950 to 2005, with a separate wartime block for 1939 to 1945 and nothing in between. The site stops in 2005 because the project stopped. And it covers what BBC News chose to broadcast, which is a different set from what happened.

Why do some dates in history look so much busier than others?

Because those edges are not evenly distributed, and three further effects stack on top.

Deliberate dates. Many anniversaries are anniversaries by arrangement. Treaties are signed on days chosen for their symbolism, ceremonies are scheduled, independence is declared on a date somebody picked. Those cluster, and the round-number rule above then amplifies them. Earthquakes get no such help.

Working days. This one is an inference rather than a sourced finding, but legislatures, courts and exchanges generate dated records on weekdays and almost none on public holidays, so a chronology built from official sources should thin out in predictable places.

Loss. On 10 January 1921 a fire in the Commerce Department building in Washington destroyed most of the 1890 United States census. Schedules for over 6,160 individuals survive, from ten states and the District of Columbia, on the National Archives' own count. A decade of American household detail is simply gone, and the fire is now the story instead.

Which calendar is a date in history even measured in?

Not always the one you are reading it in, which is why some dates come with two answers.

The Julian calendar assumed an average year of 365.25 days. The Royal Observatory at Greenwich puts the real tropical year at 365.24237 days. The gap is tiny and relentless, and over centuries the calendar slid away from the seasons. Pope Gregory XIII issued the correction in 1582.

Britain waited another 170 years. The Calendar (New Style) Act 1750 did two separate things that both matter to anyone reading an old date. It deleted eleven dates, so that 2 September 1752 was followed directly by 14 September. And it moved the start of the legal year from 25 March to 1 January, which means that under the old reckoning a date in February 1700 sat at the end of that year rather than the beginning.

So an English event before September 1752 has an Old Style date and a New Style date, they differ by eleven days, and the year can differ by one as well. Neither is wrong. They answer slightly different questions, which is worth knowing the next time an anniversary hands you a date that does not match the one you half-remember.

Time zones add a smaller version of the same trouble: one instant, two possible dates.

What never makes it into a day in history at all?

Anything nobody wrote down, kept, and later found again. Those are three filters, and the editorial rules above are a fourth sitting on top of them.

Most human lives generated no dated record at all. Where records existed they had to survive fire, flood, war, damp and the ordinary decision that a shelf was needed for something else. Then somebody had to catalogue them, and somebody had to digitise them, before an editor could cite them. The documents that survive but resist reading are their own category, and a few of them have outlasted every attempt to date or decipher them.

Each filter has the same bias: towards states with continuous bureaucracies, towards cities with newspapers, towards the recent, and towards archives digitised early. That is why Wikipedia's guideline has to tell its editors to look beyond the Anglosphere on purpose. Left alone, the pool does not.

None of that makes the format worthless. It makes it a sample rather than a census, which is a fine thing to enjoy as long as you know which one you are holding.

Fancy a different kind of dated puzzle? The daily cryptogram sets one quotation a day, and the daily hub holds the rest of the one-a-day set.

Frequently asked questions

Is there a central authority for "on this day" lists?

No, and there never has been. Nothing registers anniversaries, adjudicates between rival dates or defines what counts as historic. Each publication keeps its own file under its own rules, and several of those rules are about the file rather than about history: an observance can enter Chase's because an organisation asked, and an event can be left out of a Wikipedia listing because the article behind it is a stub. Two lists disagreeing about one date is the normal case, not an error by either.

Why do different sites disagree about what happened on a date?

Mostly because they draw from different collections and apply different rules, and partly because the date can be genuinely ambiguous. Old Style and New Style dates for English events before 1752 differ by eleven days. Events near midnight fall on either side of it depending on the time zone used. And where a source records only a month or a year, a compiler has to decide whether to place the item at all.

Does an "on this day" entry have to be important?

Not by itself. Wikipedia's guideline states plainly that selection is "NOT solely based on what is 'the most important or significant'", and lists article quality and topic mix alongside it. Round-number anniversaries are preferred, so a two-hundredth anniversary can beat a more consequential event that happens to be turning 187 this year. Importance is one input among several, not the whole test.

What is the oldest event that can honestly be given a day?

This one is reasoning from the rules above rather than a sourced finding. A day-precise date needs a calendar plus a surviving record that uses it, and both conditions get scarce fast the further back you go. Astronomical events are the exception, because the sky is predictable and an observation can be recalculated. Ordinary political and personal events usually leave no such handle, which is why so many ancient dates are given as a year, a season or a reign.

Did people really lose eleven days in 1752?

They lost eleven dates, which is not quite the same thing. The Calendar (New Style) Act 1750 struck 3 to 13 September 1752 out of the calendar so that 2 September was followed by 14 September, bringing Britain into line with the continent. The Act deliberately protected rents, annuities and lease terms so that nobody paid for time that had been deleted, which suggests the drafters expected the argument that followed.

Sources

• The World Almanac for 1876, "Record of Events", page 9 onward, digitised at the Internet Archive: archive.org/details/worldalmanac1876newy. The quoted entries for 1 January, 12 March and 20 March are from its chronology of 1875. The same catalogue record gives the publisher as Press Publishing, New York.
• Wikipedia, "Wikipedia:Selected anniversaries": en.wikipedia.org/wiki/Wikipedia:Selected_anniversaries. The project's own published guideline, and therefore the primary source for its own process. Source of every quoted rule here: the "NOT solely based on what is 'the most important or significant'" line, the five-events-a-day norm, the anniversary multiples, the no-stubs and inline-citation requirements, and the "more than just the Anglosphere" instruction.
• BBC ON THIS DAY, "About This Site": news.bbc.co.uk/onthisday/bsp/about_this_site.stm. Source of the 1950-2005 and 1939-45 coverage windows and the description of the archive's contents. The legacy host is served over plain HTTP.
• AnnArbor.com, profile of William D. Chase at ninety: annarbor.com. Source of the 1957 founding, Chase's job at the Flint Journal, his brother Harrison in Florida, the 32-page first edition, the 364 entries, the dollar price, the 2,000-copy print run, the eleven-dollar advertisement in the Saturday Review of Literature, and the 752-page, 12,000-entry figures current at the time it was written.
• Calendar (New Style) Act 1750, 24 Geo 2 c. 23, at the UK legislation service: legislation.gov.uk. Source of the eleven deleted dates, the move of New Year's Day from 25 March to 1 January, and the protection of rents, annuities and lease terms.
• Royal Museums Greenwich, "Which years are leap years and can you have leap seconds?": rmg.co.uk. Source of the Julian average of 365.25 days against a tropical year of 365.24237 days, and of the 1582 reform.
• US National Archives, "The 1890 Census": archives.gov/research/census/1890. Source of the 10 January 1921 fire and the over 6,160 people surviving in the general population schedules.