972Unsolved Historical Ciphers
p ⊓ H 9   εt y a k ┘ 8   l □ 3 9   p ⊥   9 ⊓ a
that whereas I sent … to the same

Paris → Elizabeth I · letter cipher with code signs; margin word-sign code; Scottish symbol cipher · 1559, 1568

Throckmorton to Elizabeth I, 10 July 1559, the marginal code, and John Wood to Cecil, 1568

Nicholas Throckmorton
fromNicholas Throckmorton
Elizabeth I
toElizabeth I

British Library, Add MS 4136 f. 32 (DECODE R2988) and f. 33 (R2989), in the Forbes Papers. Tomokiyo lists both as holding a cipher “yet unidentified”: a half-page of marginal signs on f. 32 and a line by John Wood to Cecil, 6 September 1568, on f. 33.

Throckmorton’s own ciphered letter of 10 July 1559 on f. 32, which is not printed in Forbes or the Calendar, was decrypted at about 85–90% with Tomokiyo’s key for his Cipher 1 and seven corrections. The two unidentified ciphers were measured and closed without a decryption: neither can be decrypted from its own text alone.

Daniel Bourdeau · posted · updated

Method: read with known key · Extent: partial (30% of the cipher)

Summary. The catalogue item was the two ciphers on these leaves that Satoshi Tomokiyo could not identify. f. 32 carries Throckmorton’s letter to the Queen of 10 July 1559, in his “Cipher 1”. Tomokiyo says that cipher fits it; no decipherment of the text was found in print (Forbes, CSP Foreign). With his key and seven corrections it decrypts as follows: Throckmorton sends a gentleman to report the French King’s state by word of mouth, asks the Queen to give him credit and reward his service, and signs off “at Paris the tenth of Julii”. Lines 1–4 are secure, lines 5–15 about 85%, and twelve code signs are open. Under it, the margin is a separate text dated “… 8 … 1559”. Its 222 groups have the type and once-only counts of English words, so it is a word-sign code or shorthand, and 51% of its tokens are signs used only once. It cannot be decrypted without its key or the 8 July despatches it probably copies (TNA SP 70/5). John Wood’s line on f. 33 has 39 signs; 92 unrelated “English” decryptions fit it better than real text, so it has no unique solution. It needs the Hatfield original. Extent: the 1559 letter decrypted in part; both unidentified ciphers closed without a decryption.

01 The leaves

Four lines of Throckmorton's cipher: small letter-like signs, digits and boxes
f. 32, the first four lines of the 10 July 1559 letter in Throckmorton’s Cipher 1. Image: British Library, Add MS 4136 f. 32, via DECODE R2988.

f. 32 (old 9) is Throckmorton’s holograph letter to Elizabeth, signed, 16 lines of cipher in his first alphabet. Below and beside the subscription are about 12 lines of small cursive signs, a date line, and a three-line block marked “*” that keys to a star in line 13 of the letter. f. 33 (old 10) is Throckmorton to Cecil, 8 August 1559 (deciphered by Tomokiyo, Cipher 3; Forbes p. 195). At its foot is one line docketed “Mr John Wod to Secretary Cecil, 6 Sept 1568” with a traced signature.

02 The letter of 10 July 1559

Key: Tomokiyo’s reconstruction of Cipher 1 (DECODE R2988 attachment), with these corrections: x and υ = i; y = f or h; z = m; ρ = d or t; ∫t and a flat-topped 6 = w; the looped Q = be; ξ and dotted q = your Majesty; a doubled sign and ζ inside a word are nulls. Code words are in square brackets; [..] marks undecrypted signs.

It may please [your Majesty] to be advertised that whereas I sent [letters] to the same by Maister Gray of the eight of [this] [month] [at] five of the clock [in] the afternoon, and [by] Carew [letters] of the same date written [at] midnight touching the [French King]’s state, I judge [your Majesty] will be desirous to be further enformed thereof. [..] convenient to put in to writing the whole discourse of [..] [which is known?] to [my] knowledge; of considering the present [..], [I judged it best?] to dispatch [this] gentleman, to whom [..] enforme the same [..] of the state [..] length, to whom it [may please your Majesty] to give credit. [..] Since [his] coming on [this] side [he hath] done [the] diverse and sundry ways [good] and painful service, [whereby he hath] deserved to be well considered and rewarded; beseeching [your Majesty] to be much [..] and gracious [lord] to him when [your Majesty] perceive [his] service [..] accepted, whereof I [have had] experience, as I shall be able to declare more plainly* and particularly to [..] when [it] shall please the same to command me to attend [..]. And [so I] humbly beseech [God] to preserve [your Majesty in] health, wealth and [all] felicity. [At] Paris the tenth of Julii.

Grades (README Conventions): lines 1–4 H; lines 5–15 C where the key gives the letters, M for the bracketed guesses. About 470 of 550 signs in lines 5–15 are decrypted with confidence. The two despatches of the 8th that it mentions are CSP Foreign I nos. 947 and 950; the bearer is not named in the decrypted part.

03 The marginal text on f. 32

Lower half of the leaf: twelve lines of small cursive signs with boxed groups, a date line with 1559, a starred block, and Throckmorton's signature
f. 32, lower half: the marginal text, its date line “… 8 … 1559” and the starred block. Image: British Library, Add MS 4136 f. 32 (lower half), via DECODE R2988.

Transcribed sign by sign (margin_transcription.txt, first pass): about 405 signs in 222 groups, about 29 base shapes (22 frequent) and about 80 shape-and-dot combinations. Groups average 1.9 signs; 106 are single signs. One boxed group occurs four times and is probably a name. Written left to right, with “/” and “//” as clause marks; not Hebrew, not mirrored or inverted Latin.

A word code. Taking each group as a word: 222 tokens give 141 types and 113 once-only types. The same measure on 222-word stretches of Bain’s calendar gives a median of 140 and 107 (margin_stats.py). Letter-for-letter English would give groups of about 4.3 signs with few single letters. The signs change value with the position of their dots, like the dotted-letter word codes of Thomas Smith (1563) and the Cecil–Norris table. 51% of the tokens are signs used once, which the ciphertext alone cannot determine, so no ciphertext-only method can reach half of the text.

What it probably is. The date line reads “… 8 … 1559”, and the letter says Throckmorton wrote twice on the 8th. The margin is likely his copy of the 8 July despatches (CSP Foreign I nos. 947, 950: Lord Grey’s ransom, the French King’s fever, forces for Scotland, Ruy Gomez, Arran at Geneva, Anne du Bourg). The calendar gives only abstracts, and the originals (SP 70/5) are only in the subscription State Papers Online.

04 John Wood to Cecil, 6 September 1568 (f. 33)

39 signs, 16 distinct, in the symbol family of Moray’s cipher to Wood of July 1568 (wood_line.txt). Bain, CSP Scotland II no. 804, prints the letter from the Hatfield original and marks one phrase as cipher: “and says he must neidis haif it be on meinis or uthir”. An exhaustive alignment (letters, nulls, syllable signs, reversed, all substrings) fails. HMC Salisbury I has no entry for the letter.

No unique solution. A solver (simulated annealing) fitted a one-letter-per-sign key against quadgram statistics of Bain’s calendar from 200 random starts; 185 end above the median score of real 39-letter text, and 92 differ from each other in more than a third of their letters, all gibberish (nonunique.py). The line is undetermined by its own text; only the Hatfield original, with whatever decipherment it carries, can settle it.

05 What remains open

06 Sources

Transcriptions, decryptions and scripts: wod1568/ in the repository.