Richelieu's Cipher to Monsieur de Rancé

01 The documents

BnF Français 3829 (formerly Béthune 9323) is a volume of original letters from 1629–1630, mostly from Louis XIII, Richelieu and Gaston d'Orléans to the Queen Mother, Marie de Médicis. Items 37 and 38 are letters from Richelieu "à monsieur de Rancé" — Denis Bouthillier, sieur de Rancé, councillor of state and secretary to the Queen Mother, and Richelieu's trusted man in her household. Item 37 is dated Saint-Privat, 7 July 1629; item 38 is undated (Avenel places it in mid-July). Both were written days after the Peace of Alès ended the Huguenot war, while the King and the Cardinal were touring the submitting towns of Languedoc.

Sensitive words and names are replaced by two-digit numbers, some carrying a diacritic; the rest is in clear. The transcription used here is the one published by S. Tomokiyo (richelieu1629.txt); the manuscript images are not freely available online.

02 The cipher system

A homophonic substitution with a small nomenclature, typical of French chancery practice under Louis XIII.

2.1 Letters (10–28): the 19-letter alphabet in scrambled order

No symbol for x, y or z was observed; i/j and u/v are not distinguished, as usual in 1629.

2.2 Vowel homophones (29–38): a e i o u twice over

So e can be written 27, 30 or 35; a 19, 29, 34 or 39; and so on. The regular a-e-i-o-u cycle makes the table easy to memorise. 33 does not occur in these two letters but is confirmed as u by Avenel's "ne piquer" (22 37 13 21 28 33 30 16) in a letter of 21 March 1629.

2.3 The diacritic: a doubling mark

A tilde or stroke over a figure doubles the letter: ~14 = ll, ~22 = nn, ~25 = ss, ~30 = ee — as in d'elle, bonne, donner, Brassac, ambassade, maistresse, amenée.

2.4 Word codes and nomenclature (40 and above)

51, 52, 58, 66, 83, 84 and 92 occur in these two letters; 54, 59 and 60 come from Avenel's footnotes to neighbouring letters in the same cipher.

03 Decipherment

Red = was in cipherBlue = nomenclature codePlain = in clear in the original

3.1 Letter of 7 July 1629 (f. 87) — French

[Le Roi] est fort en colère de l'action de Boneuil. Il remet à [la Reine mère] de le chasser, comme elle l'estimera à propos; si l'affaire est certaine, il mériteroit pis.

Castor voudroit bien que la [duchesse de Chevreuse] peust estre attrapée près de la frontière et amenée au bois de Vincennes, auquel cas il faudroit que personne de dehors ne la vist.

Les cabales de [la Lorraine] avec [l'Espagne] par le moyen de la [Chevreuse] et autres, sont insupportables.

La paix aura surpris ces esprits malins. À la vérité elle est miraculeuse, car elle couppe les racines du mal pour le présent et l'avenir. Anduze, Sauve, Le Vigan, Ganges, Uzès, Aimargues et Nismes ont desjà obéy avec grande joye, et chaque place a donné ses ostages pour seureté de la démolition de leurs fortifications. Il ne restera plus que le haut Languedoc, qui suivra bientost, Dieu aidant.

J'eusse volontiers servy Mr Vautier, mais l'abbaye estoit de longtemps promise à [La Valette]. S'il se rencontre autre occasion je m'y emploieray volontiers avec affection.

J'ay parlé à Marcheville de ce que Andreni a dit: il le nye tout à fait, et, en vérité, je ne croy pas qu'il voulust rien faire contre le service de [le Roi].

Le roy accorde que Mrs des Estats puissent faire une recreue de 1500 ou 2000 hommes, mais il désire que ce soit en Normandie et non en Picardie et Champagne, parce qu'il a besoin de gens en ces deux provinces-là pour s'opposer à ses ennemis, au cas qu'il y soit attaqué. On représentera, et fera-on bien considérer à ces Mrs que c'est une grâce très particulière que Sa Majesté leur faict. Et si [l'Espagne] s'en plaint, il faut dire qu'on n'a donné ceste permission que lorsqu'on a veu que, contre sa parole donnée, il a fait entrer les Allemands en Italie.

3.2 Letter of 7 July 1629 — English

[The King] is very angry about the action of Boneuil. He leaves it to [the Queen Mother] to drive him out, as she sees fit; if the matter is certain, he would deserve worse.

Castor [the King] would very much like [the Duchess of Chevreuse] to be caught near the frontier and brought to the Bois de Vincennes, in which case no one from outside must see her.

The intrigues of [Lorraine] with [Spain], through [the Duchess of Chevreuse] and others, are unbearable.

The peace will have taken these malicious minds by surprise. Truly it is miraculous, for it cuts the roots of the evil for the present and the future. Anduze, Sauve, Le Vigan, Ganges, Uzès, Aimargues and Nîmes have already submitted with great joy, and each town has given hostages as security for the demolition of its fortifications. Only upper Languedoc remains, and it will soon follow, God willing.

I would gladly have obliged Mr Vautier, but the abbey had long been promised to [Cardinal de La Valette]. If another opportunity arises I will gladly and warmly take it up.

I have spoken to Marcheville about what Andreni said; he denies it completely, and in truth I do not believe he would do anything against the service of [the King].

The King agrees that the gentlemen of the [Dutch] States may recruit 1,500 or 2,000 men, but he wants it done in Normandy, not in Picardy or Champagne, because he needs men in those two provinces to oppose his enemies should he be attacked there. It should be represented to those gentlemen, and made plain to them, that this is a very particular favour His Majesty is granting them. And if [Spain] complains of it, the answer is that the permission was granted only once it was seen that, against his given word, he had let the Germans into Italy.

3.3 Letter "pour M. de Rancé", mid-July 1629 (f. 89) — French

Nous sommes en peine de quoy la paix de [l'Angleterre] ne s'achève point; mandez-nous si l'on n'en a point de nouvelles.

[Le Roi] est d'advis que [la Reine mère] prépare une armée pour la Champagne de dix mil hommes et huit cens chevaux. Il en fait dresser l'estat avant que partir. Mr de Beauclerc est là pour faire ce qu'il faut. La reyne fera, s'il luy plaist, délivrer ces commissions dès ceste heure, si elle apprend le moindre bruict que les ennemis pensent à s'approcher de la frontière. Elle fera délivrer l'argent de la levée afin qu'on ne soit pas surpris. Quand les ennemis de l'Estat penseront que de tous costés l'on est préparé à se deffendre, ils ne seront pas hardis à entreprendre. Partant, quand mesmes on n'auroit aucun mauvais advis de la frontière, il est toujours expédient de délivrer ces commissions, mais non de donner argent qu'on n'ayt lieu de soubçon. Le tout dépendra du jugement de la reyne; mais en matière de précaution il vaut mieux trop faire que trop peu.

On a fort bien fait d'envoyer Mr d'Elbeuf en Picardie, et tous les gouverneurs. Il en faut faire autant en Champagne.

[La Reine mère] doibt, renvoiant ce gentilhomme de [la reine d'Angleterre], luy escrire des lettres obligeantes et pleines d'amour, et qui disent aux termes qu'il faut que [la Reine mère] se portera tousjours, pour l'amour d'elle ainsy qu'elle l'en a prié, à une bonne union entre [le Roi] et [l'Angleterre].

[La Reine mère] peut passer jusques là de dire qu'elle peut l'asseurer qu'outre l'évesque et les prestres qui doivent estre près d'elle, il ne voit pas qu'on ait dessein de luy donner d'autres François que ceux qui sont près d'elle; mais il faut que ce soit une lettre secrète, ou bien le dire en grande confiance à Vantelet qui aura créance par la lettre de [la Reine mère], ce que j'estime plus à propos; luy recommandant de ne le dire qu'à sa maistresse; et cependant il le dira aux autres, qui est ce qu'on doit désirer. Il est à souhaiter que cette paix se face et promptement.

J'ay proposé au roy Mr de Brassac pour l'ambassade de Rome. Vous sçaurez de [la Reine mère] si elle aura agréable cette proposition; et, au cas qu'elle l'agrée, vous irez trouver ledit Sr de Brassac, et luy direz la proposition que j'ai faite au roy de luy, en ceste occasion, et que si c'est chose qu'il désire, il peut dès ceste heure s'asseurer des meubles qui luy sont nécessaires pour cet ambassade.

3.4 Letter "pour M. de Rancé" — English

We are troubled that the peace with [England] is not being concluded; let us know whether there is any news of it.

[The King] is of the opinion that [the Queen Mother] should prepare an army for Champagne of ten thousand men and eight hundred horse. He is having the roll drawn up before leaving. Mr de Beauclerc is there to do what is needed. The Queen will, if she pleases, issue these commissions at once if she hears the slightest rumour that the enemy intends to approach the frontier. She will release the money for the levy so that we are not caught unprepared. When the enemies of the State see that we are ready to defend ourselves on every side, they will not be bold enough to attempt anything. Therefore, even without any bad news from the frontier, it is always prudent to issue the commissions — but not to hand out money until there is real cause for suspicion. Everything will depend on the Queen's judgement; but in matters of precaution it is better to do too much than too little.

It was very well done to send Mr d'Elbeuf into Picardy, and all the governors. The same must be done in Champagne.

[The Queen Mother], when sending back this gentleman from [the Queen of England], should write her letters that are gracious and full of love, saying in the proper terms that [the Queen Mother] will always work, out of love for her and as she has asked her, toward a good union between [the King] and [England].

[The Queen Mother] may go so far as to say she can assure her that, beyond the bishop and the priests who are to be near her, there is no intention of giving her any other Frenchmen than those already near her. But this must be a secret letter — or else said in strict confidence to Vantelet, who will be trusted on the strength of the letter of [the Queen Mother], which I think the better course — telling him to say it only to his mistress; and yet he will tell the others, which is exactly what one should wish. It is to be hoped that this peace is made, and promptly.

I have proposed to the King Mr de Brassac for the embassy to Rome. You will learn from [the Queen Mother] whether she approves the proposal; and if she does, you will go to the said Sieur de Brassac and tell him of the proposal I made to the King on his behalf on this occasion, and that if it is something he desires, he may from this moment count on the furnishings he will need for this embassy.

3.5 Historical context

The ciphered items are exactly the sensitive ones. The Duchesse de Chevreuse — the court's most dangerous intriguer, then in Lorraine and in contact with Spain — was to be seized at the border and shut up in the château of Vincennes, out of sight. The Anglo-French peace (signed at Susa, 24 April 1629) was stalling over Charles I's treatment of his French queen Henrietta Maria, whose Catholic household he had expelled in 1626; Richelieu wants the Queen Mother to reassure her daughter — by a "secret" letter or through the queen's maître d'hôtel Vantelet, which he fully expects to leak — that France will not press to reinstall a large French entourage beyond the agreed bishop and priests. Finally, Jean de Galard de Béarn, comte de Brassac, is proposed for the Rome embassy; he did in fact take up the post in 1630, an external check on the reading.

04 Method

1. Parsing. 421 cipher symbols, 46 distinct, in 58 runs interleaved with cleartext. Numbers above 40 were treated as whole-word codes and used as context breaks; cleartext words adjacent to cipher runs were kept as known plaintext.
2. Language model. French letter quadgram statistics were built from ~1.8 million characters of Project Gutenberg text (Sévigné, Retz, Pascal, Corneille, Dumas), normalised to a 22-letter alphabet with j→i, v→u and accents removed.
3. Search. Simulated annealing over the symbol→letter map, scored by quadgram log-probability plus a unigram-frequency penalty (needed because an unconstrained homophonic search collapses onto e/s/r/t). An early, wrong constraint — forcing 10–31 to be a bijection with the alphabet — nevertheless produced readable fragments (LA{58}, TRES, UNE, FRANC…).
4. Cribs. From those fragments: d'autres (fixing d a u t r e s and exposing 30 as a second e); d'elle … elle (revealing the doubling mark); then Brassac and l'ambassade after "Mr de" and "pour". With those fixed, every remaining letter fell out of huit cens, la paix, esprits, secrète, se face, Vincennes, frontière, Bonneuil.
5. Structure. The recovered values showed 10–28 to be a complete 19-letter alphabet and 29–38 two a-e-i-o-u cycles; 33 = u was predicted from the pattern before being confirmed.

05 Verification: Avenel, 1858

While searching for the meaning of the nomenclature codes, the letters were found already printed — with cipher numbers and footnote decipherments — in Lettres, instructions diplomatiques et papiers d'État du cardinal de Richelieu, ed. D.-L.-M. Avenel, vol. III (Paris, 1858): no. CXCIX, pp. 368–369 (f. 87) and no. CCV, pp. 381–383 (f. 89). Avenel writes (p. 262, n. 2) that he did not find the key among the Béthune manuscripts "mais nous l'avons recomposé à l'aide de documents que nous avons trouvés aux archives des Affaires étrangères".

His readings and the independent reconstruction agree in every ciphered word. Avenel's transcription also settles a handful of digits misread in the modern transcription (38 for 28 in l'évesque; 25 for 28 in souhaiter que; 35 for 25 in près d'elle; 25 for 26 in Vautier; 16 for 26 in Vantelet; 14 for 24 in de Rome; 22/11 for 23/13 in Champagne; 47 for 17 in frontière), none of which affected the alphabet. It also identifies the nomenclature: 51 le Roi, 52 la Reine mère, 58 Mme de Chevreuse, 66 La Valette, 83 l'Espagne, 84 l'Angleterre, 92 la Lorraine.

06 Try the key

07 Sources

• S. Tomokiyo, "Unsolved Historical Ciphers", cryptiana.web.fc2.com, and the transcription richelieu1629.txt; background in "French Ciphers during the Reign of Louis XIII", louisxiii.htm.
• DECRYPT project, DECODE database, records R9461 and R9462.
• BnF Archives et manuscrits, Français 3829, items 37–38.
• D.-L.-M. Avenel (ed.), Lettres, instructions diplomatiques et papiers d'État du cardinal de Richelieu, t. III, 1628–1630 (Paris: Imprimerie impériale, 1858), pp. 262, 368–369, 381–383. Scan: Internet Archive.

Code used for the analysis (Python: transcription parser, quadgram builder, annealing solver, renderer) is available on request.