At a glanceNot solved · text not obtained
- The letter
- Ferdinand II of Aragon to Garcilaso de la Vega, his ambassador in Rome, 31 August 1498. It encloses a copy of his letter to Joan Ram Escrivà at Naples on the peace with Louis XII; a cipher line at the head and ten cipher lines before the date are in a different hand and a different cipher.
- Where
- Barcelona, Arxiu de la Corona d’Aragó, Diversos, Sástago 193 (Lío B), nº 79/83; DECODE R10180, whose record calls the letter deciphered although only the body is.
- The cipher
- Homophonic symbols (Λ, ∠, π, ⧺o, 9, ʒ, β-forms), letter pairs written with a superscript that look like code groups, and capital letters for names.
- How it was tried
- Every key reachable online was compared sign by sign: Galende’s whole BRAH 9/15 file on DECODE, both Escrivà ciphers and the Cifra general. A homophonic annealer with a Spanish model of Ferdinand’s own letters was then run on the 315 signs.
- Why it resists
- No key fits, and 315 signs under about 57 homophones are too few for a ciphertext-only break: the same annealer recovers at most 29% of a synthetic control of that size.
- Still open
- Both passages. Garcilaso’s cipher letters of 3 October and 31 December 1496, with their decipherments, are in the Real Academia de la Historia (Salazar y Castro A-11) and would give the key.
- Credit
- Ivan Parisi located the passages and named the A-11 letters as the probable key (2004).
01 The letter and its two ciphers
Ivan Parisi published the ciphered correspondence of Ferdinand with Joan Ram Escrivà, his ambassador at Naples, in 2004. His document 83 is the odd one out: a letter not to Escrivà but to Garcilaso de la Vega in Rome, endorsed “A Garcilasso” and “Cifra d’Espanya de ses alteses 31 d’agost 98”. Most of it is a Copia of instructions sent to Escrivà, enciphered in Escrivà’s cipher no. 2. Parisi deciphered and printed that part: the king explains why he made peace with Louis XII and tells Escrivà to take leave of Naples.
Two passages are in a second hand and a second system:
| passage | where | length |
|---|---|---|
| head | image P5, one line and four signs above the word “Copia” (Parisi n. 71) | about 35 signs |
| postscript | image P4, the ten lines before the date, which is in clear: “xxxi de agosto de xcviij”, followed by two rubricas (Parisi n. 89) | about 280 signs |
Parisi called this the fourth variant and wrote that it “no ha sido todavía descifrado” despite his attempts. Satoshi Tomokiyo lists the postscript as having no known decipherment, as do Cryptoole (IT1498-1) and the DECODE record’s notes. No reading was found anywhere.
Ferdinand was at Zaragoza that month, for the Cortes; his daughter Isabel, queen of Portugal, died there on 24 August, the day after the birth of her son Miguel. The letter was written a week later.
02 What the transcription shows
Both passages were transcribed sign by sign from twice-enlarged crops of the DECODE images: 315 signs in 79 forms, or 57 when the superscript variants are folded into their base sign. Two water stains take out a few signs in the first and in the third and fourth lines. The DECODE images are not in the public domain and publication needs the archive’s permission, so no crop is reproduced here; the transcription is in the repository.
- Homophonic symbols. Most signs are symbols, and many are frequent: β with a ring (19), 9 (17), π (17), dotted Λ (15), dotted ∠ (15), ⧺o (12), T with a small o (12). The seven commonest cover 40% of the text, which points to several signs per letter.
- Code groups. About forty tokens are lower-case letters with a superscript and a point: aq· (7 times), et, af, as, ad, bn·, hn·, ge·, ke:. They look like the two- and three-letter code groups of the Catholic Monarchs’ ciphers (ab, eb, ib … bag, beg, bog), written with the second letter raised.
- Names. Capitals B, M, R, T and H stand alone, as name signs do in the keys of the period.
- Repeats. “π bong zh” occurs in the first and the sixth line; “ꝑ π” five times, twice in the head line; “aꝛaꝛ” four times.
03 The keys that were tested
The Real Academia de la Historia volume 9/15, which Galende Díaz described in 1994 as the Cifras de los Reyes Católicos, is on DECODE as key records R9815–R9832 and R9486. It was not known here to be online. Each alphabet was compared with the postscript’s signs.
| key | result |
|---|---|
| ff. 11–14v, cipher between Alonso de Silva and Garcilaso de la Vega (R9822) | Some letter signs agree (9, ʒ, ∝, <, ⊤), and its code is built like the superscript pairs. But Λ, π, ⧺o, T-with-o and the β-forms are not in it, and 9 (its m) occurs too often to be m. |
| f. 16, a Rome cipher of about 1495–96 (R9824) | Its name list is Roman (the pope, the College, the cardinals of San Pietro in Vincoli, Naples, Cartagena and San Giorgio, the vice-chancellor), and its capitals B, M, R, T, H coincide with the postscript’s. Its alphabet does not: none of Λ, ∠, T-with-o or zh. Galende assigned this folio to fray Juan de Mauleón. |
| ff. 7–10, the Cifra general (R9821) | different alphabet |
| ff. 1–6, 15, 17–25 (R9815–R9820, R9823, R9825–R9832, R9486) | different sign sets |
| Escrivà’s ciphers nos. 1 and 2 (Parisi 2004) | no. 2 shares some shapes (π, Λ, ∝), but applied to the postscript it gives nonsense, with z as the commonest letter |
| other ciphered letters of the period in RAH 9/7–9/8 on DECODE (R9483–R9485, R9591) | different systems |
A place-and-date crib, “en Çaragoça a” before the clear date, fits the last signs only if Λ is g, which its frequency rules out.
04 The ciphertext-only attack
A homophonic solver (simulated annealing) assigned one letter to each sign form. It was scored with a 5-gram Spanish model built from Golden Age prose and Parisi’s printed decipherments of Ferdinand’s own letters of 1495–98, with a penalty against collapsing onto vowels. Restarts gave different strings that are not Spanish, and no two agreed.
The same solver was then given a control: 315 letters of Spanish from a text outside the model, enciphered with 57 homophones, with no code groups and a perfect transcription. Six restarts recovered between 0 and 29% of the letters. At this length the method cannot succeed, so its failure on the postscript says nothing about the cipher beyond its being short.
05 What would finish it
- Images of Garcilaso’s cipher letter to Ferdinand of 3 October 1496 (RAH, Salazar y Castro A-11, no. 1067, with its decipherment no. 1076; 9/11 ff. 109–111 in Galende’s numbering). Parisi named it as the likely key.
- His despatch of 31 December 1496 (A-11 ff. 121r–125v, deciphered on f. 125), cited by Villanueva Morte and Fernández de Córdova (2025).
- D. W. McPheeters’s 1989 paper on an unknown letter of the Catholic Monarchs to Garcilaso, not seen here.
Neither Salazar letter is on DECODE or in the RAH Biblioteca Digital. With either, the alphabet and the code could be rebuilt from cipher and decipherment and applied to both passages.
06 Sources
- ACA, Diversos, Sástago 193 (Lío B), nº 79/83: DECODE R10180 (images P4 and P5).
- Ivan Parisi, “La correspondencia cifrada entre el rey Fernando el Católico y el embajador Joan Escrivà de Romaní i Ram”, Pedralbes 24 (2004), 55–116, doc. 83 and notes 13, 71, 89 (UB).
- J. C. Galende Díaz, “La escritura cifrada durante el reinado de los Reyes Católicos y Carlos V”, Cuadernos de Estudios Medievales XVIII–XIX (1994).
- BRAH 9/15 on DECODE: R9822 (ff. 11–15), R9824 (f. 16), R9821 (Cifra general).
- C. Villanueva Morte and Á. Fernández de Córdova Miralles, El embajador Claver. Diplomacia y conflicto en las «Guerras de Italia» (1495–1504) (CSIC, 2025), n. 152.
- S. Tomokiyo, Cryptiana, open problems page; Cryptoole IT1498-1.
Transcription, solver, control and notes: targets/garcilaso1498.