972Unsolved Historical Ciphers
39 16 96 100 38 97 74 22 102 75 · 73 38 97 75 83 98 20 74 · 15 72 89 11 29 81
u n s z u s e t z e · e u s e r s t e · g e f a h r

Bremen → Johan Adler Salvius, Hamburg · homophonic cipher of two-digit figures inside clear German · November–December 1631

Riksarkivet cipher letters, 1628–1633

Johan Adler Salvius
toJohan Adler Salvius

Riksarkivet, Stockholm, via DECODE R4282–R4341 (fourteen records, catalogue item 207) · catalogued as one Latin cipher, “unknown sender to unknown recipient”, 1628–1646 · in fact four unrelated groups

The key of the archbishop of Bremen’s two letters of 1631 was recovered from a contemporary decipherer’s interlinear glosses and the grid order of its values, and both letters are deciphered in full · a Latin letter cipher, a Latin numerical cipher of 1628 and a numerical letter to Amsterdam of 1632 were attempted and not solved · the other nine records were not worked

Daniel Bourdeau · posted · updated

Method: key recovered based on adjacent plaintext · Extent: partial (30% of the cipher)

Summary. The catalogue entry treated fourteen DECODE records from the Riksarkivet as one body of ciphered Latin despatches of unknown origin. Opened, they are four groups: three ciphertexts filed in the key boxes (Chifferklaver låda II 113, 114 and 133), two letters of Johann Friedrich, archbishop of Bremen, of 1631, five letters of J. J. von Rusdorff to Axel Oxenstierna of 1628, and four Swedish letters to Oxenstierna in the Oxenstierna collection (E 708), dated June 1633, not 1646 as DECODE has it. The Bremen letters are deciphered in full. They are addressed to Johan Adler Salvius, the Swedish resident at Hamburg, not to Åke Tott. A clerk deciphered about four fifths of their figure groups between the lines. Matching those glosses to the numbers showed that every letter owns four consecutive numbers, and that the bands follow the alphabet written in a six-column grid and read down the columns. That placed the letters no gloss covers, and every group is now decrypted. The three key-box ciphertexts were transcribed, attacked with a solver (simulated annealing) in ten languages, and tested against two keys from the same boxes. None was solved. The Rusdorff and E 708 letters were only surveyed. Status: partial. 2 of 14 records deciphered, 3 attempted and open, 9 not worked.

01What the fourteen records are

RecordsWhat it isSystemHere
R4330, R4331Johann Friedrich, archbishop of Bremen, to Salvius at Hamburg, 27 November and 2 December 1631; Germantwo-digit homophonic figures inside the clear text, and a few one- and two-digit name codesread
R4282 (låda II:113)Latin letter, about two pages, undated, no namesletter cipher of about 33 signs (Latin letters, 3 4 5 7 8, λ Δ φ ε α and a boxed square), Latin phrases left in clearattempted, open
R4284 (låda II:114)Latin letter of 1628, docketed with the letters to Ludwig Camerarius108 different numbers, comma-separated, Latin words in clear; a leaf of key tests beside itattempted, open
R4306 (låda II:133)letter to a “Mr …ilick” at Amsterdam, “1632?”463 numbers, mostly 30–52; index of coincidence 0.078, so one sign per letterattempted, open
R4333–R4337Rusdorff to Oxenstierna, The Hague, 1628figures inside clear German and Latin; the archive calls many of them unsolvednot worked
R4338–R4341 (E 708)Swedish letters to Oxenstierna, 8/18 and 27 June 1633two-digit letters, four-digit name codes and letter pairs; part deciphered between the lines at the timenot worked

02The Bremen key

Lines of German with runs of two-digit numbers and small interlinear glosses
R4330, the letter of 27 November 1631: figure runs inside the clear German, with the decipherer’s words above (“Gronsfeldische”, “Colonien”, “troupen”). Image: Riksarkivet, Stockholm, letter of 27 November 1631, via DECODE R4330.

Thirty-five glossed words were set against their numbers one letter to one number. Twenty-three matched in length, and the numbers for each letter came out as runs of four: 8–11 for a, 16–19 for n, 72–75 for e, 96–99 for s. The runs are not in alphabetical order. The alphabet is written in six columns (a b c d e f / g h i k l m / n o p q r s / t u w x y z) and read downwards, so the bands go a g n t, b h o u, c i p w, d k q x, e l r y, f m s z:

8 a · 12 g · 16 n · 20 t · 24 b · 28 h · 32 o · 36 u · 40 c · 44 i · 48 p · 52 w · 56 d · 60 k · 64 q · 68 x · 72 e · 76 l · 80 r · 84 y · 88 f · 92 m · 96 s · 100 z

No gloss covers k, z, m, w or y. The unglossed groups fix them: sincken, schantze, kommen and dem accord mitbegriffen, etwas and wir, weyter and Reuterey. Once they are in place, the grid predicts q and x without a single example. Small numbers stand for words: 4 = wir/uns, 5 = Herzog Christian, 7 = Herzog Georg, 32 = the Stift Minden, 59 = Salvius.

03What the letters say

The archbishop, Johann Friedrich of Holstein-Gottorp, wants Swedish help, and quickly. In the letter of 27 November he fears that the enemy will be in the Stift before relief comes, “und in die euserste gefahr kommen würden”, and asks that “mit schleunigem hulf und succurs ausgeholffen werden möchte”. His people have taken die Burg and are attacking die schantze. They cannot hold without more men.

The letter of 2 December is more urgent. The Reuterey ranges up and down the country, laying waste one place after another, and he fears an einfal by Bönninghausen und Gronsfeldischen troupen from the Westphalian border. He hopes to be “mit etlichen tausend secundiret”, or at least to get etwas fusvolckes und theils dragoner. A postscript asks that, if the garrison of Newenburg marches out on terms, his ertzstift be in dem accord mitbegriffen.

The full decipherment, line by line with the glosses, is in bremen_decoded.txt. The remaining errors are transcription slips from the phone photographs (occupireh, feldlsihen); no group is left undecrypted.

04The three ciphers not solved

R4306, to Amsterdam, has one sign per letter by its statistics, and “34 42” alone occurs 31 times. Simple and homophonic substitution were annealed over quadgram models in Dutch, German, Latin, French, Swedish, English, Italian, Spanish, Portuguese and Danish. None fits. A second transcription kept the scribe’s variant figures apart (34 and 74, 40 and a looped 4o, 30 and 70), on the chance that they are separate signs. That did not help either. It remains open whether the transcription or the model of the cipher is at fault.

R4284 of 1628 uses 108 numbers across 756 groups, with Latin words in clear. A homophonic Latin search produced Latin-sounding nonsense, and so did a search on bands of four. Beside it lies a leaf on which someone tried a letter cipher on four phrases: effusorem sanguinis, pestem patriae, praeter naturalem, disturbatorem religionis. That leaf is unused material.

R4282 is about a thousand signs of Latin letter cipher, with et qualis sit eius futurus status dubitatur, sed tamen ut res and Mittatur nobis responsum in clear. It was annealed as Latin with and without the written word divisions, and no plaintext appeared. Seventy key records from the same boxes were fetched. The two that looked closest, a Transylvanian secretary’s key (R4310) and a 1632 key whose 30 ends every word (R4296), were tested and fit neither of the numerical letters.

05Sources

Transcriptions, the decipherment, the solver and notes: riksarkivet1628/.