At a glanceNot solved · text not obtained
- The letter
- János Pápai, Ferenc Rákóczi II’s envoy to the Porte, to Rákóczi, Landor Fejérvár (Belgrade), 6 March 1710, signed Pápai János; in clear Hungarian apart from six short passages.
- Where
- Magyar Nemzeti Levéltár, G15 Caps. H Fasc. 226, pp. 31–33; DECODE R731 (catalogue item 46).
- The cipher
- A graphic alphabet of perhaps 25 signs (∂, π, an m-like sign, q, ÷, =, “11”, with colons and dots), about 330 signs in all: three passages on p. 1 and three on pp. 2–3. Not the numeric key Pápai used in his other despatches.
- How it was tried
- Transcribed at 1.8–1.9× zoom and attacked with a swap-annealing substitution solver scored by a model of Pápai’s own deciphered Hungarian. The solver recovers a control text of the same lengths exactly.
- Why it resists
- On R731 it finds no Hungarian under any reading of the transcription (as transcribed, without dots, without the swash stroke, reversed). Either signs are merged in the transcription or the dots carry letters, and the 2448-pixel camera images cannot settle which.
- What would finish it
- Better photographs of pp. 31–33, or the sign alphabet itself.
- Other ciphers
- Pápai’s eleven numeric despatches of 1706–10, in the key issued to him (DECODE R580), are deciphered on their own page.
01 The letter
R731 is written in clear except the six passages, and DECODE has no transcription of them. The pages carry no interlinear decipherment. The key copy R581 has a line of signs of the same family in its top-left corner, which the DECODE transcriber noted as “symbols which do not occur in the key itself”; read the right way up it is a line of cipher, not an alphabet, so it gives no key. The images carry the archive’s notice on DECODE that publishing them needs its permission, so no crop is shown here.
02 What was tried
The passages were transcribed into R731_signs.txt: about 330 signs, 15 sign types plus dot clusters. The transcription is tentative. The low swash stroke may be a pen link rather than a sign, and the dots above and below the bars (÷, ∴, ::) were collapsed into one token, although they may be vowel marks.
solve731.py is a swap-annealing substitution solver with a 4-gram model trained on Pápai’s deciphered letters. On a control, Pápai’s own text cut to the same line lengths and enciphered at random, it recovers the text exactly (−8.96 per 4-gram, against −8.87 for real text). On R731 it reaches only about −10.1 under every reading tried. So R731 is not a simple Hungarian substitution on this transcription. A recurring group has the letter pattern of Pasa read backwards, but it did not lead to a key.
Kálmán Benda’s edition of Pápai’s Turkish diaries (1963) is not online; whether it prints these passages is not known.
03 What would finish it
A sharper photograph of the three pages, to fix the sign boundaries and the dots, and the solver run again; or the sign alphabet from the Rákóczi chancery papers. Attempted, not solved.
04 Sources
- DECODE record R731 (Magyar Nemzeti Levéltár, G15 Caps. H Fasc. 226, pp. 31–33; images by login); key copies R580, R581, R452.
- Kálmán Benda (ed.), Pápai János törökországi naplói (Budapest, 1963), not consulted.
Transcription and solver: targets/papai1710/ in the repository. Split on 14 October 2026 from the page of Pápai’s numeric despatches, which are in another key.