Bifid cipher

Leave the square key empty for a plain A to Z square, or type a word to mix it. Period sets how many letters are worked together. At 0 the whole message is one block, and a period of 5 or 7 is common for longer messages.

Decoding needs the same key and the same period, or the digits pair up with the wrong neighbours and you get nonsense back. Letters only, in capitals, and J is read as I.

Put the alphabet in a 5x5 square, with I and J sharing a cell, and give every letter its row and column number. Write each letter's row under it, then each letter's column under that.

Now read the numbers across, all the rows first and then all the columns, in one long line. Take that line two digits at a time, and look each pair up in the square as a row and a column. Those are the cipher letters. That's all.

Wikipedia's example uses the square BGWKZ QPNDS IOAXE FCLUM THYVR. There, FLEE AT ONCE gives the line 44 33 35 32 43 13 55 31 23 25 and comes out as UAEOLWRINS. Type that square into the key box to see it here.

Félix Delastelle, a French cryptographer who studied the subject as a pastime, invented the bifid. Wikipedia's article on the cipher dates it to around 1901, while his biography says he first presented it in 1895. It combines fractionation, splitting each letter into pieces, with transposition, moving those pieces around.

Because every cipher letter mixes digits from two different message letters, simple letter counts tell an attacker very little. He went on to stretch the same idea from a square to a cube in the trifid cipher.