The History of Sudoku: From Euler to Your Newspaper

Ask most people where Sudoku comes from and they will say Japan. It is a reasonable guess — the name is Japanese, and Japan is where the puzzle got its name and its polish — but the modern puzzle was invented in the United States by a retired architect from Indiana, built on mathematical ground that was surveyed in Europe two centuries earlier. The full story runs through Basel, Seoul, Paris, Indianapolis, Tokyo, Hong Kong, and London, and it is worth telling properly.

Latin squares: the mathematical ancestor

Strip away Sudoku's 3×3 boxes and you are left with a Latin square: an n×n grid in which every symbol appears exactly once in each row and each column. The Swiss mathematician Leonhard Euler studied these systematically in his 1782 paper Recherches sur une nouvelle espèce de quarrés magiques (“Investigations on a new type of magic square”), and his name has been attached to them ever since. Euler was interested in when two Latin squares could be overlaid so that every ordered pair of symbols appeared exactly once — the famous “36 officers problem” — not in recreational puzzles, but the row-and-column constraint at Sudoku's core is his territory.

Euler was not even first. The Korean mathematician Choi Seok-jeong published a pair of orthogonal Latin squares of order nine in his book Gusuryak decades before Euler's paper — a 9×9 arrangement, no less, though without Sudoku's box constraint. Combinatorics has deep roots.

A Latin square is a weaker object than a Sudoku grid. There are roughly 5.5 × 10²⁷ Latin squares of order nine, but only about 6.67 × 10²¹ valid Sudoku grids — the box constraint eliminates all but roughly one in a million of them. If you want the details, we work through the counting in The Mathematics of Sudoku.

The French newspaper puzzles of the 1890s

For about a decade either side of 1900, French newspapers ran number-placement puzzles that came startlingly close to modern Sudoku. Le Siècle published partially completed 9×9 magic squares to be finished by the reader. Its rival La France refined the idea in 1895 into a puzzle whose solution had each of the digits 1–9 exactly once in every row and column. These puzzles even tended to have the digits fall neatly into the nine 3×3 regions — though the setters never stated the box rule explicitly, and the puzzles used arithmetic (broken diagonals summing correctly) rather than pure logic. The French puzzles disappeared around the time of the First World War, apparently without descendants. Sudoku would have to be invented again from scratch.

Howard Garns and Number Place (1979)

The puzzle we play today first appeared in the May 1979 issue of Dell Pencil Puzzles and Word Games under the name Number Place. Dell did not credit individual constructors, so the inventor's identity had to be worked out later by detective work: puzzle historians — Will Shortz, the New York Times crossword editor, prominent among them — noticed that Howard Garns, a retired architect and freelance puzzle constructor from Indiana, appeared on the contributor list of every Dell issue that contained a Number Place and was absent from every issue that lacked one.

Garns was 74 when the first Number Place ran. He took Euler's row-and-column idea and added the crucial third constraint — the nine 3×3 boxes — and the convention that a puzzle should be solvable by logic alone. He died in Indianapolis in 1989, before his puzzle conquered the world, and almost certainly never knew what it would become. There is no evidence he ever earned more than Dell's standard contributor fee for it.

Nikoli and the Japanese refinement (1984)

In 1984 the Japanese puzzle publisher Nikoli spotted Number Place in the Dell magazines and introduced it to Japanese readers in its magazine Monthly Nikolist. The company's president, Maki Kaji, gave it the name Sūji wa dokushin ni kagiru — roughly “the digits must remain single,” a pun on unmarried people — which was soon abbreviated to Sudoku (数独, “single numbers”).

Nikoli did more than rename the puzzle. Two editorial rules, introduced in 1986, turned a curiosity into a phenomenon in Japan:

• No more than 32 givens. Fewer clues meant the puzzle demanded real deduction rather than bookkeeping.
• Rotationally symmetric clue patterns. The givens form a pattern that looks the same turned 180° — an aesthetic borrowed from crossword grids that made the puzzles beautiful as objects.

Sudoku thrived in Japan partly because the Japanese writing system makes crosswords hard to construct, leaving an appetite for language-independent pencil puzzles. Interestingly, “Sudoku” is a Nikoli trademark in Japan, so rival Japanese publishers mostly call the puzzle nanpure (Number Place) — the puzzle is known by its Japanese name everywhere except Japan. Maki Kaji, often called “the godfather of Sudoku,” remained a beloved figure in the puzzle world until his death in August 2021.

Wayne Gould and the global boom (1997–2005)

The final act begins in a Tokyo bookshop in 1997, when Wayne Gould, a recently retired Hong Kong judge from New Zealand, picked up a partly completed Sudoku book on a whim. Gould spent the next six years writing a computer program, Pop-Up Sudoku, that could generate puzzles with graded difficulty in effectively unlimited supply — the thing a daily newspaper needs.

In 2004 he walked into the offices of The Times in London and offered the puzzles for free, betting he could make money on books and software instead. The Times printed its first Sudoku on 12 November 2004. Within months every British national paper had one; within a year the puzzle was global. 2005 was the year of the boom: bestseller lists filled with Sudoku books, the first World Sudoku Championship was organized for March 2006 in Lucca, Italy, and the puzzle settled permanently into the daily routine of millions of people alongside the crossword.

What each era contributed

Why the history matters to solvers

The history explains features of the puzzle that solvers take for granted. The promise that a puzzle is solvable by logic alone comes from Garns and Dell. The expectation of a single unique solution — which some advanced techniques exploit directly, as we explain in Why Every Proper Sudoku Has Exactly One Solution — became an ironclad convention in the Nikoli era. And the graded difficulty labels on every modern puzzle are Gould's legacy, for better or worse; we look at what those labels actually measure in What Makes a Sudoku Hard?

The techniques themselves — Naked Singles, Hidden Pairs, X-Wings, Swordfish — are younger than you might think. Most were named and codified between 2000 and 2006 by online solving communities racing to crack the hardest published puzzles. That story belongs to the individual technique guides, and every one of them can be practiced on real puzzles right here: choose a technique and try it.