Mirabeau's Cipher

Mirabeau's Cipher

The Count of Mirabeau, an early leader of the French Revolution, used a cipher in his correspondence with his mistress, Sophie, Marquise de Monnier (Wikipedia) before he played a major role in politics.

The cipher represents letters of the alphabet with two digits stacked on top of each other. The specimen below (where I could not figure out how the first word can be deciphered as "Gien") shows Arabic figures were also represented by such stacked digits, with "0" on top (see "deux"). There are additional symbols for alternative representation of the letters and short words.

Mirabeau and Sophie

Mirabeau (1749-1791) met Sophie (1754-1789) in 1775 in Pontarlier in Franche-Comté when he was supposed to be locked up in the Château de Joux by an arrangement made by his father to protect him from creditors but was actually allowed much freedom (Wikipedia in French). Both unhappy with their respective marriage, they fell in love and eloped, much to the scandal for both their families (Delon). They were caught in Amsterdam. After giving birth to a daughter (only to live less than three years), Sophie was locked up in the convent of Saintes-Claires in Gien in 1778. Mirabeau was put in the dungeon of Vincennes.

Mirabeau was released in December 1780 and managed to see Sophie in the convent in 1781. But Sophie was gradually disillusioned of the passion she believed they had been fostering through correspondence (Delon). Sophie was freed by the death of her elderly husband in 1783 but remained near the convent and committed suicide in 1789.

Mirabeau was reconciled with his father, who began to see political talent in him. In 1786, he was employed in a secret diplomatic mission to Berlin (see citations in my blogpost). In the early stages of the French Revolution, although he was a count himself, he led the commoners before his premature death in 1791.

Correspondence

In the dungeon of Vincennes, Mirabeau was allowed by Police Lieutenant Lenoir (or Le Noir) to correspond with Sophie via an unofficial channel through Boucher, chief clerk of the Secret Service (premier commis du Secret) called "Bon Ange" (or "BA" for short) in the correspondence. Because the original letters had to be returned to Lenoir (who in turn acted under M. de Maurepas), they were in the police archives. (Université de Tours (2018) p.140, 145; Cottin (1903), p.lx)

Communication through Boucher was not a perfect solution for the lovers, because the letters were read by Boucher. After Sophie arrived in Gien, they tried writing with lemon juice, but Boucher discovered this in March 1779 and made them promise never to repeat it. So, from 18 June 1779, they resorted to conducting secret correspondence besides the official channel by bribing the guards. (Cottin (1903) p.lxii-lxiii, lxv, lxxiii).

Cipher

To secure secrecy further, a cipher was introduced. It was around March 1780 that Sophie received a first cipher alphabet from Mirabeau (BnF NAF 13189, f.134, may be an earlier use if the dating "15 fevrier - 1780?" is confirmed. The note on f.36 for 1776 says "aux trois quarts indechiffrable" but this must simply refer to the handwriting.). Soon, however, judging it was too difficult for its weakness, Mirabeau had another cipher (including monosyllables) composed by a friend, Dubut de la Tagnerette, son of the Postmaster General (Cottin (1903) p.lxv).

At first, Sophie used the cipher sparingly but soon complied with Mirabeau's request for "beaucoup de chiffres". (Cottin (1903) p.lxvii)

Publication of the Letters

Mirabeau's letters during his imprisonment in Vincennes (1777-1780) were published in Lettres originales de Mirabeau, écrites du Donjon de Vincennes (1792) by Pierre Manuel, who had access to the letters because he became the municipal administrator of the police (Wikipedia) after storming of the Bastille.

Manuel had believed he identified the key but, a century later, Paul Cottin, a historian and editor, found that it did not serve to read the "hiéroglyphes" in the letters from Sophie to Mirabeau. It was the illustrious codebreaker Étienne Bazeries who reconstructed the key (Cottin (1903) p.lxix). Bazeries says in Les chiffres secrets dévoilés (1901) that publication of the letters had been delayed presumably because of their "pornographic" content (Bazeries (1901), p.19; one word I randomly picked in the middle of f.141v of NAF 13189 deciphers as "clitoris").

Cottin published Sophie's letters in Sophie de Monnier et Mirabeau d'après leur correspondance secrète inédite: 1775-1789 (1903) with a long introduction, though bawdy contents were silently suppressed (Delon). (Bawdy or not, it can be seen in the above image that about ten lines of cleartext after the ciphertext are omitted in the published text.) The cipher table is on p.lxv (see above).

Manuscript Sources

Sources of documents related to Mirabeau are partially listed at Portail Biblissima and those related to Sophie are partially listed at Portail Biblissima. Besides these, BnF NAF 18312 includes one letter by Mirabeau to Sophie, dated 2 February 1780 (f.35; no cipher). But these do not seem to include the bulk of the letters published.

Most of the original letters were destroyed, though copies were preserved as of Cottin's work. Cottin had about 250 letters from June 1776 to June 1781, either original or copies, of which only a few were included in Manuel's compilation and of which Cottin included about a third or a quarter in his publication. (Cottin (1903) p.lxii-lxiii, lxv, lxxiii)

BnF NAF 13189 ("Lettres de Sophie Richard de Ruffey, marquise de Monnier, à Gabriel Riquetti de Mirabeau, 1776-1781") seems to be the largest extant source, but I'm not sure of the exact nature of this volume. (Cottin says on p.lxv that his file contained no letter from Gien to Mirabeau before 13 March 1780. This is almost consistent with the content of NAF 13189, in which there is no letter between f.128 (25 July 1777, not from Sophie) and f.136 (30 March 1780), if we do not count f.133-f.135, for which dating to February 1780 may be questioned. But why "13 March" rather than 30 March?) Considering that Cottin (1903) includes many letters not in this volume, he must have had other sources. Archivists' advice is needed in this regard.

Still, many of the letters in Cottin (1903) can be found in this volume. For example:

Apart from the nature of BnF NAF 13189, there are some questions to be answered.

What was the Contribution of Dubut de la Tagnerette?

Cottin says after Sophie received a first cipher alphabet around March 1790, Mirabeau considered it was too difficult while easy to break, and had another cipher (including monosyllables) composed by a friend, Dubut de la Tagnerette, son of the Postmaster General (Cottin (1903) p.lxv). But it is not clear from the manuscripts what change was brought about after the first introduction of the cipher. The basic cipher used in the earliest period in BnF NAF 13189 f.137 (30 March 1780) appears to be the same as the one used in the above image from December 1780.

The term "monosyllable" is used by Cottin to mean a monosyllable code such as Lol to represent Mme de Lamballe (Cottin (1903), p.159, lxx; used in, e.g., Sophie's letter from 24 July 1780). But initials such as C. for Briançon had been in use even before the introduction of cipher (instances from 1776 are in Cottin (1903), p.45, etc.).

The same term is used by Manuel to mean a monosyllable plaintext word represented by a special symbol (see his key below).

Is it possible that the entire cipher with stacked digits and other symbols for monosyllables was prepared by Dubut de la Tagnerette, which replaced a simpler cipher? Although use of such an earlier naive cipher is not found in NAF 13189, the letters in the volume may not be complete.

Another possibility is that Cottin's wording "monosyllable" is misleading and Dubut de la Tagnerette advised introduction of the mixed alphabet (see the next subsection).

The episode about Dubut de la Tagnerette is probably based on the letters in Cottin's hand. It will be best to identify the original records but reading the context of letters around 22 April 1780 may hopefully give some hint. On this day, Mirabeau (the letter on f.141 is addressed to "Madame"; the letter is not in Manuel's publication, vol.4) seems to complain about Sophie's misuse of cipher ("tu met toujours les uns pour les autres" (f.141, 17th line from the bottom)). A new cipher may have been sent shortly before this because Sophie's letter from the same date mentions "ton alphabeth de chiffre nouveau" (NAF 13189, f.145, line 5-6).

Given the letter from 22 April has a larger variety of symbols than the letter from 30 March, Dubut de la Tagnerette might have just expanded the vocabulary. Even if Mirabeau was not an expert in cipher, however, it is wondered whether he needed a friend's help for that.

What Was Bazeries's Contribution?

According to Cottin, the key in Manuel's publication was incorrect and did not allow reading the "hiéroglyphes" and Bazeries reconstructed the key for him.

One can see that Bazeries' key for two stacked digits is already seen in Manuel's key.

Manuel provides two keys: "Alphabet naturel" (regular alphabet in normal arrangement) and "Alphabet transposé" (mixed alphabet). He demonstrates the use with the regular alphabet. For example, E is the fifth letter in the first group, so it is enciphered with "5 over 1". Bazeries' key for the digit cipher does not match this but it generally fits Manuel's key for the mixed alphabet. (Manuel considered J was not in the alphabet.)

Cottin must have provided Bazeries with Manuel's key when he enlisted the codebreaker's help. His wording "hiéroglyphes" suggests he was aware that there was no problem in Manuel's key for the digit cipher. But if it is only the graphic symbols Bazeries identified, it does not seem to be much of an achievement. Did Bazeries break the digit cipher from scratch, perhaps because he didn't examine Manuel's key closely, taking Cottin's words to mean that it was totally incorrect?

Regarding this, Bazeries himself says as follows in verifying a similar cipher in a book by Boetzel and O'Keenan.

By saying "This system was revealed to us by the decipherment of some authentic letters of the Marquise Sophie de Monnier to her famous lover M. de Mirabeau.", he is not exactly claiming he broke the cipher. Perhaps he sorted out the reading of some graphic symbols and possibly also explained the working of the digit cipher to Cottin.

Was Not Cipher for Secret Channel?

Manuel (1792) published the key (see above), which he dared because there was no longer Bastille (vol.1, p.xxxii). It is wondered whether the cipher was used in any of those letters published therein (presumably sent via the official channel). If so, did not Boucher find and make an issue of it, as he did when he found writing in lemon juice?

Scribble, Shorthand, or Cipher?

BnF NAF 13189 includes sheets with an archivist's note. Some of such sheets look cryptic. It does not look like a cipher or cleartext. Can it be a shorthand?

References

Pierre Manuel (ed.) (1792), Lettres originales de Mirabeau, écrites du Donjon de Vincennes (vol.1, vol.2, vol.2, vol.4, Wikipedia in French)

Paul Cottin (ed.) (1903), Sophie de Monnier et Mirabeau d'après leur correspondance secrète inédite: 1775-1789 (Google)

Étienne Bazries (1902), Les chiffres secrets dévoilés (1901) (Wikimedia)

Delon, M. (2010). "Lettres d'amour, lettres de combat Sophie de Monnier et Mirabeau (1775-1789)", Revue de la BNF, 35(2), 18-24 (DOI, cairn)

Université de Tours (2018), "Mirabeau sous le sceau du secret : l'écriture épistolaire à l'épreuve de la surveillance pénitentiaire", Lumen, 37, 135-150 (DOI)

©2026 S.Tomokiyo
First posted on 2026. Last modified on 11 July 2026.
Articles on Historical Cryptography