A Short History of the Crossword from 1913 to Your Phone

On 21 December 1913 the New York World printed a diamond-shaped grid under the heading FUN, labelled word-cross, made by an editor named Arthur Wynne who needed something to fill a page. It had no black squares, a hollow centre, and thirty-odd clues.

A typesetting error a few weeks later reversed the name to cross-word, and it stuck. Everything that follows is downstream of a filler puzzle and a printing mistake.

The 1920s craze

The format spread slowly for a decade and then all at once. In 1924 two young publishers, having failed to find anything more promising, released a collection of the World's puzzles. It sold in enormous numbers and effectively founded Simon & Schuster.

What followed was a genuine public mania, and a moral panic to match. Newspapers ran editorials about wasted hours. Libraries reported dictionary shortages. A railway line was said to have put dictionaries in its carriages. Contemporary accounts read almost exactly like every subsequent panic about a new leisure format.

Every mass entertainment format gets one round of editorials predicting the collapse of concentration. The crossword got its round in 1924.

Standardisation

The rules we now treat as fundamental were not present at the start. Wynne's grid had no blocks and no symmetry. Through the late 1920s and 1930s a convention settled: a rectangular grid, rotational symmetry, every white square checked in both directions, a minimum entry length of three, and interlock throughout.

Those constraints were not handed down; they emerged because they made puzzles better and easier to edit at scale. Once a newspaper is publishing daily, consistency is worth more than novelty, and consistency requires rules.

The New York Times held out for years, dismissing the format as a fad, and only began publishing a crossword in 1942 - reportedly as a distraction during wartime. It has since become the reference standard for the American style, which is a very particular kind of late-arriving dominance.

The British divergence

Britain took the format somewhere else entirely. By the early 1930s British setters were writing clues built on wordplay rather than definition, and the cryptic crossword emerged as a distinct discipline with its own conventions, its own indicator vocabulary and its own fiercely defended notion of fairness.

The grids diverged too. British grids use far more black squares, allow unchecked letters, and typically have lower interlock - which is exactly what you need when every clue is a small puzzle in itself. Two traditions, same ancestor, incompatible instincts. Our introduction to cryptic clues covers the mechanics.

The computer changes construction

Until the 1990s, filling a grid was entirely manual. Constructors kept card indexes and word books and filled by hand, which is why older grids often have block patterns that look conservative to a modern eye - open grids were simply too expensive to attempt.

Construction software and machine-readable word lists changed the economics. Wide-open themeless grids with long stacks became routine, word count fell, and the standard for acceptable fill rose sharply, because the tooling made clean fill achievable rather than heroic. Our guide to the modern construction stack covers where that landed.

Digital solving

The last shift is the most recent and the least settled. Solving moved to apps, which brought timers, streaks, autocheck and leaderboards, and turned a private leisure activity into something with a public score.

Two consequences matter for the craft. First, difficulty became measurable at scale for the first time, which is a genuinely new thing in a century-old format. Second, some construction devices work worse on a screen than on paper - the rebus square being the obvious casualty.

The format also fragmented. Mini crosswords, themed daily mobile games and puzzle apps built around a single mechanic now reach far more people than the broadsheet grid does, and they operate under much tighter vocabulary constraints. A themed mobile crossword with its daily solutions written up and a long-running syndicated newspaper daily are a useful pairing if you want to see how differently the two ends of the market now behave.

What has not changed

Remarkably little, structurally. A 1935 grid and a 2026 grid are recognisably the same object, solvable with the same instincts. The fill is cleaner, the clues are livelier, the cultural references have turned over several times, and the underlying contract - every letter checked, every clue fair, exactly one intended solution - is identical.

That durability is unusual. Most popular formats from the 1920s are gone. This one survived because the constraints that emerged in its first fifteen years happened to be very good ones, arrived at by editors solving a practical problem rather than by anybody designing a game.

If you want to read further, the biography of the format's inventor is short and worth it, the general crossword entry is a good survey, and the grid-level historical record at XWord Info lets you watch the conventions change year by year. For a broader look at how puzzle formats have branched over the same period, our own history of the logic grid puzzle covers a parallel and much shorter lineage, and the Word Ways archive is the standing journal of record for recreational linguistics generally.