The prince de Conti’s ciphered mémoires, March 1649

Summary. Three ciphered mémoires of the prince de Conti, generalissimo of the Paris Fronde, for his envoys with Archduke Leopold Wilhelm in March 1649. The catalogue of new targets listed the mémoires of 26 and 27 March (nos. 41, 42) as two ciphers without a decipherment, and assumed the third (no. 43), which has one, would be the crib. The two parts have different outcomes.

Nos. 41 and 42: known key, three small corrections. George Lasry deciphered both in 2023 and published the key: a letter cipher with homophones and polyphonic symbols, which he calls “poly-homophonic”. Here the leaves were transcribed afresh from the Gallica images and decrypted with his key. The text agrees with his; three small corrections are listed separately in section 02.
No. 43: a different system, and its unglossed opening. It is a number code of letters and syllables with a contemporary decipherment between the lines, which starts only at the thirteenth line. The key was rebuilt from about 3,300 glossed groups and used to decipher the opening, 264 groups: the address to Laigue and the purpose of the whole mémoire. No other decipherment of these lines is known. The whole mémoire was then edited: Conti to Laigue, 5 March 1649, the day after the Rueil conference opened, explaining the conference away to the Archduke.

01 The documents

BnF Français 3854 (Gallica btv1b52520094g) is a recueil of Fronde papers from the prince de Conti’s side, January to April 1649. It holds his instructions to his envoys, the marquis de Noirmoutier and Geoffroy de Laigue, the Archduke’s Spanish answers, and three pieces in cipher that Omont describes as follows:

No. 41 (f. 113r–v): “Mémoire, avec chiffre, du prince de Conti. A Paris, ce 26 mars 1649”.
No. 42 (f. 115r): the same, “ce 27 mars 1649”, signed Armand de Bourbon.
No. 43 (ff. 117r–120r): “Mémoire, avec chiffre et déchiffrement, du prince de Conti, pour un de ses envoyés auprès de l’archiduc”, undated.

The catalogue’s “verify first” was to confirm that the three share one cipher. They do not. Nos. 41 and 42 are written in lower-case letters mixed with clear words. No. 43 is written in numbers, with a clerk’s letters and syllables above most of them.

BnF fr. 3854 f. 113r, the head of no. 41: A Paris ce 26 Mars 1649, and the first lines of the letter cipher mixed with clear words such as Monsieur, choses, comme, Il sera — No. 41, f. 113r: “A Paris ce 26 Mars 1649”, then the letter cipher inline with clear words. *Jcctg nle Monsieur de nrgmtild* is “Affin que Monsieur de Noirmo[u]stier”. Image: Bibliothèque nationale de France, fr. 3854 f. 113r, via Gallica (btv1b52520094g).

02 Nos. 41 and 42: Lasry’s decipherment, and three corrections

Prior art. A web search before any work found George Lasry, “Armand de Bourbon’s Poly-Homophonic Cipher – 1649”, HistoCrypt 2023. Lasry deciphered both letters, printed their text with a translation, and described the system.

How it works. Each plaintext letter is written as a cipher letter. A, L and R each have two symbols (homophones). Five symbols each stand for two to four letters (polyphony): a = t/p/a/b, e = e/g, i = c/z, m = s/d/x, n = m/q.
The key. There is no nomenclator. The key’s second row spells J’aime c’est un grand mal, a Boesset air of 1642.
The catalogue entry. His paper cites the volume as “fr. 3584”, a typo, which is why the catalogue did not match it.

Decrypted again here. All three pages were transcribed symbol by symbol from the full-resolution images. Because five symbols are polyphonic, decryption means choosing a letter for each of them: a search over the possible letters, scored with a French quadgram model and then a period word list, made the choice. The text agrees with Lasry’s throughout, apart from the three points below.

Corrections to Lasry’s text of nos. 41 and 42

f. 113r: dans lequel temps les choses ne pouvant pas. The cipher has ge = NE, which the printed text drops.
No. 42: ordonne à ses députez de se joindre à ceux de son Altesse le prince de Conti pour insister à demander l’éloignement du cardinal Mazarin. The words are in the cipher (arldtgmtmaed) and in Lasry’s own figure, but not in his clean text.
f. 113v: Bretaigne is enciphered with the L symbol in second place, a slip of the encipherer’s.

“Monsieur de Tenerande”, with whom Longueville is to treat, decrypts the same under the key; the person is unidentified. “Dix” or “six mil hommes” cannot be settled, because one symbol serves both d and s.

Depuis la lettre que je vous escrivi hier j’ay cru vous debvoir advertir qu’on a donné aujourd’huy arrest au parlement qui ordonne à ses députez de se joindre à ceux de son Altesse le prince de Conti pour insister à demander l’éloignement du cardinal Mazarin. On ne doute plus après cela de la rupture … Ne concluez rien pourtant que vous n’ayez le dernier advis, mais si elle arrive, mettez-vous en estat d’en faciliter les suittes. No. 42, 27 March 1649

03 No. 43: a number code, and twelve lines without a gloss

BnF fr. 3854 f. 117r, the first thirteen lines of no. 43: rows of numbers with clear words (quil, principalement, bonnes intentions, jay jugé apropos de vous en, que vous puissies, par les, nous sommes tousiours dans), and the first interlinear gloss po ur p r o cu at the foot — No. 43, f. 117r. Twelve lines of numbers and clear words with nothing written above them; the clerk’s interlinear decipherment starts only at the foot (*po ur p r o cu*, “pour procu[rer]”). Image: Bibliothèque nationale de France, fr. 3854 f. 117r, via Gallica (btv1b52520094g).

The clerk who deciphered no. 43 wrote a letter or syllable above almost every number from f. 117r line 13 to the end of f. 120r. He wrote nothing over the first twelve lines. Those 264 groups carry the address and the purpose of the mémoire.

The key, rebuilt from his glosses. Number/gloss pairs were transcribed page by page (by model sub-agents), and the tallies (key43.txt) show a regular design:

Four interleaved single-letter alphabets: 3–25 run a→z, 28–50 run z→a, 51–73 run a→z again, and about 80–98 run down.
Syllables in blocks of five, in vowel order: ca ce ci co cu = 256–260, and so on through the consonants, with a second, homophonic series in the gaps.
Underlining a number reverses its syllable: 385 = ra, underlined 385 = ar.
Code words: 451 la Reyne, 457 le cardinal, 458 le parlement, 472 l’archiduc, 482 France. The 500s are dates.
Nulls: the numbers from about 99 to 250, which are never glossed.

The opening, deciphered. Of the 264 groups:

220 take the value the same number has in the glosses.
24 follow from the alphabets or are nulls.
20 are syllables of the second series, fixed from context.

The first draft was checked against the rest of the mémoire. Nine of the context and structural values were later found glossed on ff. 117r–120r, all as predicted: 279 de, 364 la, 354 ne (underlined, en), 443 te, 445 su, 37 o, 43 h, 86 n, and 36 p. None was contradicted.

Pour Monsieur de Legue. Parce que nous ne doutons que le cardinal ne donne une mauvaise interprétation à toutes ses actions, et qu’il ne face principalement ses efforts pour mettre l’archiduc dans la défiance de nos bonnes intentions sur le sujet de la conférence que [le] parlement a accordée, j’ay jugé à propos de vous en faire instruire, afin que vous puissiés destourner les mauvais effets des artifices du cardinal, et faire comprendre à l’archiduc que ce pourparler n’a esté causé et résolu que par les retardemens que l’archiduc apporte à nous secourir, et que nous sommes tousiours dans nos mesmes desseins … No. 43, f. 117r, lines 1–12, not glossed by the clerk; deciphered here

04 What no. 43 says

The last line of f. 120r is glossed à Paris le 5me mars 1649. That is the day after the Rueil conference opened between the court and the parlement. Conti writes to Laigue, who is with Archduke Leopold Wilhelm, to keep the Spanish alliance alive while the Frondeurs negotiate with Mazarin’s side.

Why the conference happened. The parlement laid the Archduke’s envoy’s proposals (dom Joseph) before the Queen. Mazarin, who saw that real peace would ruin him, proposed a conference to drag things out and to make the Archduke believe the Frondeurs were giving way. “À quoy l’on a employé Vautorte.” Conti would have refused it if the Archduke’s help had been ready and Paris not starving.
What it bought. No suspension of arms. Saint-Germain lets in a hundred muids of corn a day from Corbeil for as long as the conference lasts, which buys time, “mais fort peu, en vous attendant”.
Why it cannot succeed. The deputies are “tous gens de bien, dans le zèle de la paix générale et dans une haine irréconciliable pour le ministériat du cardinal”. They cannot give way on Mazarin, whom the parlement has condemned. The generals stayed away on purpose, so as to be free to break the talks at will.
The promise. “Dès le moment qu’il entrera en France, cette conférence se rompra”, and a truer one will be opened with the Archduke’s own deputies.
The plea. “Il n’y a plus un seul moment à perdre pour nous secourir.” Laigue is to press for help with all the Archduke’s forces. Even if the Archduke will not march without a treaty, nothing is to stand in the way of relief.

The complete edited text of the mémoire is in targets/conti1649/READING43.md.

Where it fits. The three mémoires frame the Frondeurs’ double game. On 5 March, Conti promises the Archduke that the conference is only a way to buy time. On 26–27 March, he tells Noirmoutier that the talks are about to break on Normandy and Mazarin, and that he should be ready. The Peace of Rueil was signed on 11 March, in between. The parlement’s arrest of 27 March shows the generals still pressing the Mazarin article after the peace.

05 What remains uncertain

The opening of no. 43. Ten of the opening’s context values (277, 314, 321, 358, 390, 398, 400, 407, 439, 442) never occur in the glossed text, so they rest on sense alone. None of them changes a sentence.
Four short places in no. 43’s glosses are faint, blotted or struck, and are left as gaps in the edition: f. 117v l. 18, f. 118v ll. 5–8, f. 119v l. 14 and f. 120r l. 0. The clerk wrote something at f. 119v l. 14 and struck it through, apparently “duc de Lorraine”.
No. 41: Tenerande is unidentified, and “dix” or “six mil hommes” is open.

06 Sources

BnF Français 3854, nos. 41–43, ff. 113–120. Gallica btv1b52520094g, canvases 237–251; description in BnF Archives et manuscrits.
George Lasry, “Armand de Bourbon’s Poly-Homophonic Cipher – 1649”, Proceedings of HistoCrypt 2023, Linköping Electronic Conference Proceedings. ecp.ep.liu.se.
Code and data: targets/conti1649/.
- Nos. 41–42: transcription (transcription.txt), decoder (decode2.py) and reading (READING.md).
- No. 43: pairs and key in n43/; edition in READING43.md.