Four letters catalogued as Ferenc Rákóczi II’s, 1706–1711

01 The documents and the keys

R639 is a one-part numerical nomenclator, headed De Monsieur de Bonac et Graffei. Codes 10–120 are letters and endings, 121–460 syllables and common words, and 461–560 titles, peoples and places. The odd numbers from 97 to 119 are nulls, and so are the groups of 500 and above that open a line. Transcribing the sheet from its image corrected DECODE's copy of it. The S and T columns have 45 and 47 written over other figures, and DECODE transcribed them as 95 and 96. The Z column has 54, 55, 110. The numbering jumps from 172 (Da) to 175 (de), yet the letters use 173 and 174 constantly as “de” (“le Roy 173 Pologne”, “Palatin 174 Belz”). 561 and 562, beyond the end of the table, stand for “nt”.

The new transcriptions corrected about 40% of DECODE's lines. They restored a line DECODE had dropped in both R912 and R852, removed repeated runs, and split merged groups.

02 R852: Dantzig, 22 January 1706

Monseigneur, j’ay receu la lettre que V. A. m’a fait l’honneur de m’escrire le 26. X^bre. Je feray passer au S^r Groffey celles qui estoient pour luy … V. A. peut luy donner directement les ordres, outre que je vous respons encore une fois de sa fidelité … J’ay veu une ou deux fois un imprimé latin intitulé Mercurius hungaricus. Il seroit tres important pour le service de V. A. qu’on imprimast regulierement toutes les semaines une feuille de cet imprimé … il n’y a point de gazetier allemand ny hollandois qui ne detruise à son ayse le Roy^me d’Hongrie … On croioit encore à Berlin qu’il n’y avoit plus de Hongrois depuis l’action.

The ciphered page is followed by its clear copy, transcribed here from the image, and by a short clear note offering to keep a register of M. de Bonnac's letters and bills of exchange. The writer is at Danzig and speaks of Groffey in the third person. The cipher's letters and syllables agree with R639, but its word codes do not: 144 is de and 254 is la, where R639 has beaucoup and jusque. It is an earlier issue of the table.

03 R922: Dantzig, 28 October

Monseigneur, je me suis privé pendant quelque temps de l’honneur d’escrire à Vostre Altesse … fait à Madame la Palatine de Belz … une copie authentique de cette vente au sieur Kray … autant qu’il a esté possible par mon credit et par mes avances … pour joindre le Roy de Pologne … l’avantage de luy continuer mes services à la Cour de Pologne … je suis avec tout le respect possible, Monseigneur.

The letter concerns a sale for the Palatine of Belz (Elżbieta Sieniawska), orders for carbines, and the sieur Kray. On page 2 it turns to money and to someone who is a prisoner.

A contemporary hand deciphered only the first two lines, between the lines: the date and “je me suis privé pendant quelque temps de l’honneur d’escrire à Vostre Altesse”, 37 of the 722 groups. Everything after them was deciphered here with R639 alone.

04 R912: “au quartier”, 5 November

J’ay eu l’honneur de marquer à Vostre Altesse … le Roy Stanislas … le Palatin de Belz … le grand Général … la pacification … une nouvelle élection en Pologne … le Czar … escrite au maréchal Rehnskiöld … Monsieur le Palatin de Russie … avec un profond respect et un zèle parfait.

A survey of Polish and Swedish politics that includes the Rehnskiöld affair of R902. It also covers a regiment asked of Rákóczi, which King Stanisław and the Germans hold against him, and Tököly.

What was deciphered at the time. A contemporary hand wrote the plaintext between the lines of page 1 over every line except the end of one and the two below it (659 of the page’s 713 groups), over the first four lines and the start of the fifth on page 2, and over the first line and the start of the second on page 3: about 780 of the letter’s 1,630 groups. DECODE’s transcription records these interlinear lines. The opening, the Palatine of Belz, the grand général, the pacification, the new election, the Tsar and the letter to Rehnskiöld quoted above all lie under it. The rest of pages 2 and 3 carries no decipherment on the leaves and was deciphered here with R639 alone: the Palatine of Ruthenia, the regiment, Stanisław and the Germans, Tököly and the closing formula. So were the three lines of page 1 that the decipherer left.

Those three lines finish a sentence about the King of Sweden, “ce Prince s’est voulu réserver un prétexte de rester cet hiver en Pologne pour ne pas s’éloigner de l’Allemagne”, where the interlinear stops (added 30 September 2026):

… ni perdre de veue les affaires generales de l’Europe. Aussy publie-t-on comme une chose presque certaine qu’il n’entrera en Moschovie que la campagne qui vient.

‘… nor lose sight of the general affairs of Europe. So it is given out as an almost certain thing that he [Charles XII] will not enter Muscovy until the coming campaign.’ He did invade Russia only in the campaign of 1708. Of the 54 groups, 44 read directly with R639 (H), four are fillers or nulls (589, 619; 115, 97), and six need comment:

• publie: the group is 327 = pu. Its last figure has the open top of this hand’s 7 (as in 257 and 297), not the round bowl of its 9 (349, 149). It had been transcribed 329 = pas (M).
• une: the group is 87 = U. Its first figure closes in a loop at the foot, like the 8 of 87 in “un homme” higher on the page, not the open hooked 5. It had been transcribed 57 = Et, here and by DECODE (M).
• veue: 427 426, both ve in the key. The key has no separate v (its alphabet runs T, U, X), and a line higher the contemporary decipherer wrote ue over 427, so the pair reads ve-ue (M).
• presque: 343 350, présent que in the key. The writer used 343 for the pré- of prétexte two lines higher, where the decipherer also wrote pre (M).
• campagne: 58 = ns on the image and in the key, read as a slip for 88 = M (M).
• vient: 534 46, Vienne and T. The place code is used for its sound, as 266 leur gives the l’Eur- of l’Europe in the same line (M).

The first transcription also had 509 (France) at the end of the last line. The image has 589, a filler, as DECODE reads it, so R912 has 1,585 keyed groups, not 1,586.

05 R483: Rákóczi, Lwów, 27 June 1711

This letter is signed in autograph “Franciscus Princeps” and dated in clear “Leopoli die 27. Junii Anno 1711”. It is addressed to a trusted envoy on his way to Holland. Cipher and clear alternate within the lines, and the clerk wrote the decipherment above every ciphered stretch. Rákóczi sends a copy of the “Tractatus sic dictae Pacis Carolianae”, the Peace of Szatmár, which he calls a fraud worked out in Vienna. He will not recognise a king proclaimed in the counties outside a free diet, contrary to the 1687 article of Pozsony:

nos Regem extra dietam promulgatum, non deposito [juramento], agnoscere non posse sine … ruina [libertatum nostrarum].

He appeals to the Tsar and the Polish Republic to mediate. He offers to come to a free diet and recognise a lawfully crowned king if the confederates recover their estates. He calls the Allies' assurances a “diversio armorum” and asks for the war to go on. He also reports Pálffy's cannon before “arcem nostram Munkatsiensem”, the confederation in Poland and the Muscovite advance. A clear postscript concerns mathematical instruments ordered from Mangold.

The key. Each cipher line was aligned with the Latin above it by an alignment algorithm (hard expectation-maximisation), with each code taking up to four letters. The result is a syllabic cipher with homophones. Single letters have several values (190 and 430 s; 140 and 380 n; 60 and 300 e; 20 and 260 a), open syllables have their own codes (257 re, 428 ti, 68 se, 362 de, 96 no, 457 ri, 92 ce, 152 co), and a few words and names are coded whole. Of 224 codes, 156 occur at least twice. Decoding the letter with this key reproduces 70% of the clerk's letters. Where the clerk's handwriting is doubtful, the cipher usually settles the word.

06 What remains uncertain

• R912 and R922: 62 groups not deciphered. They are line-opening and line-closing fillers, R912's closing block of small figures (probably a date and a name), and values outside the table: 93 (×3), 89 (×3), 9 (×2). 173/174 = de and 561/562 = nt are inferred from use (C).
• The year 1707 for R912 and R922 rests on content (C). R852's 1706 is written on the page.
• Writer: unsigned. R852 and R922 come from Danzig, and R852 names Groffey in the third person (M).
• R852's variant word codes have not been tabulated. Its clear copy gives the text.
• R483: 68 of the rebuilt codes occur only once (C). A line-by-line edited Latin text has not been made yet.

07 Sources

• Magyar Nemzeti Levéltár (MNL OL), G15 Caps. C. Fasc. 39 and P237 Festetics 10. d. 2. Images are on DECODE (login) and are not reproduced, as the archive requires.
• DECODE R922 (images 4911–4912), R852 (4706–4708), R912 (4877–4879), R483 (3666–3671).
• Key: DECODE R639 (G15 Caps. C. Fasc. 44/08, image 3927), first matched here for R902.

Transcriptions, keys, aligner and decipherments: targets/rakoczi1704/ in the repository.