The Great Cipher

The Great Cipher (Grand Chiffre) was the secret code of the French kings, designed in the seventeenth century by the cryptographers Antoine Rossignol and his son Bonaventure. It was so strong for its time that after the system lapsed, nobody could read the surviving letters for about two hundred years. A French army officer, Étienne Bazeries, finally broke it in the 1890s after three years of work, and the letters that came out of it set off a famous argument about the Man in the Iron Mask.

How it worked

The Great Cipher was a nomenclator: a book of numbers, each standing for something. The core set had 587 numbers. Most stood for syllables of French, some for single letters, and a few for whole words or names. A message was a long string of numbers, and the recipient looked them up in the same book.

The system beat the usual attack on a substitution cipher, counting letters, for several reasons. The symbols were syllables, not letters, so no 26-letter frequency table applied, and a table of 587 numbers is far bigger than an alphabet. Frequent syllables, like frequent letters, had more than one number, which broke up the repeats. Some numbers were nulls, meant to be ignored, and some were traps, such as a number that canceled the one before it. The codebooks were secret and they changed, so no outsider could get started by reading a few known letters. Almost nobody outside the French court's cipher clerks knew how to use it.

The Rossignols

Antoine Rossignol became famous in 1626, when he broke a message intercepted near Réalmont, where Huguenot forces were besieged. Reading it led to their surrender and made his name. He and his son Bonaventure later developed the Great Cipher, which French monarchs used for generations. The French army was still using a modified form in the Peninsular War, until 1811.

After that the key was lost, and many documents in the French archives could no longer be read. The cipher picked up a reputation as one that could never be broken. That's a bold claim for any cipher, and this one held only as long as nobody tried hard enough.

Bazeries breaks it

Commandant Étienne Bazeries, a cryptanalyst in the French army, took on the problem around 1890 and worked on it for about three years. His first insight was that the numbers stood for syllables, not letters or whole words, which is the one thing you need to know to start. He then looked for repeated groups of numbers and guessed that one repeated string meant les ennemis, "the enemies", a phrase that must come up often in letters about war. With a few syllables fixed, the rest followed, number by number, and by about 1893 he could read the documents. It's the same method used against every substitution cipher on this site: look for patterns, guess a probable word, and let the guess spread. The simple substitution cipher page describes it for letters. The Great Cipher is the same idea scaled up to 587 symbols and a language of syllables.

The Man in the Iron Mask

One of the letters Bazeries read, from 1691, passes between Louis XIV and his general Nicolas de Catinat. It concerns another general, Vivien de Bulonde, who had failed to take Cuneo in Piedmont and been disgraced, and it recommended that he be imprisoned at the fortress of Pignerol. Pignerol was also the prison of the mysterious prisoner who became famous as the Man in the Iron Mask. Bazeries argued that the prisoner was Bulonde, on the strength of two groups of numbers, 330 and 309, which appear only once in the surviving documents. He read them as the word masque and a full stop, which would make the king order Bulonde to be kept masked.

The historian David Kahn, in his history of codebreaking, pointed out that no known version of the cipher from that period contained the word masque, and that the two numbers appear only once, so nothing confirms the reading. The identity of the real prisoner is still debated, and this letter doesn't settle it. It's a good example of a danger in codebreaking. When a solution works for most of a text, a doubtful reading of the unique leftover groups can steer you to the conclusion that fits the story you wanted.

What it teaches

• Big codebooks are still substitutions. Replacing letters with syllables raises the cost of an attack but doesn't change what kind of cipher it is. Patterns, and a good guess at one word, break it.
• Secrecy of the book cuts both ways. If the book is lost, the history is lost. The book cipher has the same problem.
• A reputation for being unbroken lasts only until someone tries. The cipher looked safe for two centuries because it had fallen out of use, and nobody had a reason to break it until an army officer wanted to know what the French court had said.

Modern ciphers avoid all of this by using a published algorithm with a short secret key and relying on it to survive public attack (AES-GCM, explained).