I have been in correspondence with Christian Boyer from France. He has recently written an article for "Pour La Science" (the French eidition of "Scientific American") on the subject of Sudoku's history.

We know about Euler's involvement with Latin Squares, dating to his paper of 1782. We know that the first known appearance of a Sudoku puzzle in the USA was in 1979. Did nothing happen between 1782 and 1979?

It is sometimes claimed, on the strength of the 1979 first known appearance, that the puzzle was invented then. However, the first known appearance of something is not proof of the time of its invention. It certainly constitutes circumstantial evidence of the time of invention, but it is not proof in itself. Perhaps I am influenced in this by my years as a Judge.

Christian Boyer has uncovered a series of puzzles which appeared in French newspapers in the 1890s. French dailies and magazines published a variety of games that featured the ingredients of Sudoku, such as:

* 9x9 grid with 3x3 boxes;

* blanks to fill with numbers;

* the digits 1 through 9 in each row, column and even box.

Not all of those ingredients come together in every puzzle, and they shared the Daily Puzzle slot with other kinds of puzzles. However, Christian illustrates some remarkable puzzles, including one (dating to July 6, 1895) which -

* if solved like a Sudoku puzzle produces 2 solutions;

* but which if solved with the setter's additional rule that the broken diagonals must add up to 45, produces 1 solution.

The way the puzzle was printed at the time, it did not have the 3x3 boxes appearing with heavy outlines. It seems the setter was approaching it more as a magic square than anything else. However, coincidentally or not, the puzzle does comply with the basic Sudoku rules, including the requirement that each 3x3 box contain the digits 1 through 9. However, the single correct solution does allow repetition of the digits in broken diagonals.

Christian's research may have uncovered the "missing link" - or at least one of the missing links - between Euler and Sudoku.

You can find Pour La Science at www.pourlascience.com

- Wayne