The Great Cipher
Le Grand Chiffre of the Rossignols — the nomenclator that kept Louis XIV's secrets for two centuries
Why This Matters
The Great Cipher is one of history's most successful codes. For over two centuries, French diplomatic and military correspondence sealed by the Rossignols sat in archives — completely unreadable. When the codebook was lost after the Rossignols' deaths, even the French government could no longer decipher its own messages. It took until 1893 for the cryptanalyst Étienne Bazeries to crack it, finally letting historians read the secret correspondence of the Sun King.
Antoine Rossignol (1600–1682) was France's first true cryptanalyst. His talent was discovered at the 1626 siege of Réalmont, where he broke a Huguenot ciphered letter overnight — and the rebels surrendered the next day. Cardinal Richelieu brought him to court, and Antoine, joined by his son Bonaventure, designed a new cipher so strong it became the official cipher of the French crown.
The Rossignols' code book contained roughly 587 numbers, each standing for a syllable, a common word, a single letter, or — critically — nothing at all. Some numbers were pure decoys (nulls); at least one number meant "ignore the previous letter". After Bonaventure's grandson Antoine-Bonaventure died in 1751, the codebook itself vanished, and the Rossignol papers in the French archives became unreadable artifacts.
In 1890 the historian Victor Gendron passed a stack of these encrypted letters to Étienne Bazeries (see Exhibit 21; full biography in Hall X · Codebreakers), already famous for breaking diplomatic ciphers. Bazeries spent three years on it. His breakthrough: noticing that one common number was followed only by certain other numbers — and that interpreting it as "delete the previous letter" suddenly produced French words. Among the messages he read was finally an answer to one of history's great mysteries: a 1691 letter naming a prisoner who would later become legend as "the Man in the Iron Mask."
Unlike a substitution cipher (one symbol per letter), the Great Cipher is a nomenclator — a coded vocabulary. The codebook lists numbers paired with syllables:
124 → "les" 301 → "roi" 412 → null (ignore) 87 → "que" 055 → "Q" 333 → null (ignore) 156 → "men" 702 → "MAJESTÉ" 891 → "DELETE PREVIOUS" 209 → "ent" 044 → "X" ...
To encrypt, the clerk worked through the message picking the longest matching codeword from the book — RE, EN, TRE, etc. — falling back to single letters only when no syllable matched. Occasionally the clerk would sprinkle in null codes as decoys. To decrypt, you needed the codebook in front of you. Without it, the message was just a long list of three-digit numbers.
Bazeries assumed the codebook used syllables, not just letters. He counted the most frequent codewords across hundreds of letters, hypothesizing that the most common one represented les — by far the most common French syllable. From there he tested combinations against known historical context (place names, royal titles, dates of known events).
For two centuries every attacker assumed each codeword stood for something. Bazeries noticed that one common number, no matter how he interpreted it, produced gibberish — but if he treated it as a command to erase the preceding letter, French words emerged. This single trick — borrowed from the medieval Italian tradition of cifre dolose ("malicious figures") — was what kept the cipher unreadable for so long.
A modern hill-climbing solver, given enough ciphertext and a French syllable-frequency model, can recover most of the codebook in minutes. The cipher's true defense was the labor and time required by 19th-century pen-and-paper analysis, not its mathematical strength.
Among the letters Bazeries finally read in 1893 was a dispatch from war minister Louvois to General Catinat dated 1691, discussing the imprisonment of one "general Vivien de Bulonde" who had cowardly lifted the siege of Cuneo. The order: arrest him, take him to the fortress of Pignerol, and "have him conducted to the prison at night with a mask on his face."
Bazeries believed this was the famous Iron Mask. Most modern historians disagree — Bulonde was alive and free again by 1708 — but the discovery showed the world that the Great Cipher had finally fallen, and that its protected secrets had been historically significant ones.
| Concept from the Great Cipher | Modern Evolution |
|---|---|
| Variable-length codewords (syllables, not letters) | Compression + encryption: encoding redundancy makes plaintext smaller and statistical attacks harder |
| Null tokens / decoys | Padding schemes (PKCS#7, OAEP) and chaff-and-winnow techniques |
| Hidden control codes ("delete previous") | Format-aware deception in steganography; honey encryption that produces plausible plaintext for wrong keys |
| Lost codebook = lost messages forever | Key management is the hard part: a strong cipher with lost keys is indistinguishable from no cipher at all |
| Security through operator skill, not algorithm | Modern principle (Kerckhoffs): the algorithm should be public; only the key must be secret |
| Exhibit | 38 of 40 |
| Era | Baroque · ~1626 |
| Security (then) | Effectively unbreakable for 200 years |
| Security (today) | Broken by hill-climbing |
| Inventors | Antoine & Bonaventure Rossignol |
| Codebook size | ~587 numbered codewords |
| Broken By | Étienne Bazeries · 1893 |
| Famous use | Louis XIV's diplomatic & military mail |
The historical Rossignol codebook (587 entries) is lost in detail. This demo uses a teaching-scale codebook — 50 common digrams + 26 letters + 6 nulls + 1 "delete previous" trap (83 codewords) — deterministically generated from the Codebook Seed.
Try encoding the same message twice: notice how the output changes because common syllables have multiple codes and nulls are sprinkled in randomly. Decoding always recovers the original (provided the seed matches).