A cipher used in a letter from Feuquière (Wikipedia, Wikipedia, his memoires (1741)) to General Catinat, dated 15 January 1691, presented in another article was broken by Daniel Bourdeau in September 2026 (see his github). While he has been making full use of AI in breaking many unsolved ciphers on my list (he told me he solved as many as 26 in a few days), automatic codebreaking did not work for this particular case.
The letter is printed in Mémoires et correspondance du maréchal de Catinat (1819), vol.2 (Google), p.283. The cipher used in this letter is different from the one used in the seven other letters printed therein addressed to Catinat by Louvois or Louis XIV. The latter cipher is the famous great cipher broken by Commandant Bazeries (see my articles: "Commandant Bazeries' Codebreaking and His Candidate of the Man in the Iron Mask", "Specimens of Louis XIV's Great Cipher (1691)" and Bourdeau's, "Louis XIV and Louvois to Catinat, July-September 1691").
After unsuccessful attempts by AI (recorded in his github), Bourdeau found, in vol.1, p.140, another letter in cipher, dated 6 September 1690, to d'Herleville, Feuquière's predecessor as governor of Pignerol, and saw that it was in the same cipher (judging from common figure groups and bigrams/trigram etc.).
Moreover, the ciphertext is followed by a contemporary translation, though from two other letters to Catina, which can be read with Bazeries' great cipher, it was seen that the translation was paraphrased by the clerk.
Bourdeau noted repeating syllables "au", "ra", and "le" in "comment aura réussi vostre entreprise sur le chasteau de Villefranche" at the end of the translation corresponding to the ciphertext in the d'Herleville letter.
The phrase, albeit possibly paraphrased, must be in the latter half of the last paragraph in cipher, where corresponding repeated figures are found as expected.
From this, with some experience in segmentation into letters/syllables in syllabic ciphers, it is relatively straightforward to derive the correspondence:
This assignment is further supported by repeated occurrences of 21(s) and 101(c) (and 91(t) immediately before the quotation). (Fortunately, this part in the translation was not paraphrased.)
Once this much was found, the rest is an ordinary process of identifying one element after another.
When these results were applied to the Feuquière letter, words such as "heures du matin", "du vingt-quatre", "m'escrire", and "duplicata" were identified, which revealed further symbols, and so on.
When one is given the above matching pair of ciphertext and decryption, it may not seem a difficult task. But I salute the ingenuity of focusing on that phrase in the first place in the paraphrased translation (and before that finding the d'Herleville letter in vol.1 that I overlooked).
In 1690, Louvois used a grand chiffre with the highest number 587 in correspondence with Catinat, while he used a petit chiffre (the highest number is 366 in Bourdeau's key) with the governor of Pignerol. In 1691, Catinat used the same grand chiffre with the court and the same petit chiffre with the new governor of Pignerol. Although the petit chiffre has fewer entries, it shares the strength of the grand chiffre such as two-part structure (random arrangement of entries); homophonic assignments not only for letters but also syllables (e.g., DE=25/42/71); and mixed arrangement of letters, syllables, and (admittedly only a few) words.