Thirty favourites from the archive

01 Ten that made us laugh

Drunk envoys, royal gossip, catalogue clerks with the best of intentions, and one minister who could not quite work his own code.

01

1533Hungary → the Dukes of Bavariakey from adjacent plaintext

“To be spoken with in the morning, when he is sober”

DECODE filed this letter under “Augurellio”, the name on its address leaf. At the foot of two leaves it is signed, in cipher, Iorg Weinmaister: Jörg Weinmeister, a Bavarian agent in Hungary. The key was rebuilt from contemporary glosses on sibling letters in the same volume and on this one. Most of the letter is grim news, including the siege of the castle of Palota and a report of fifty thousand Turks marching from Belgrade on Buda. Then the talks on relieving Palota run into a practical problem. The magnate Török Bálint was sent to discuss the relief one evening in Vienna while well drunk, and will have to be approached again the next morning.

the reading, in English

…one evening at Vienna well drunk, was sent to talk of the relief, and is to be spoken with in the morning when he is sober.

Weinmeister 1533, 96.7% deciphered

Iorg
Weinmaister
· signed in cipher ·

02

1580Rome → the nuncio in Parisexisting decipherment

The Pope, the King of France, and the nun

This is a cipher enclosure from the papal Secretariat of State to the nuncio Dandini, with the clerk’s decipherment written under the figures. George Lasry decrypted its companion letter in 2020 with his reconstructed key. This one uses the same nomenclator with different letter values, which were rebuilt here with a solver (simulated annealing) and checked against the clerk’s text. The two agree word for word, except that the cipher has commercio where the clerk wrote concetto. What Rome wanted kept secret was the gossip about Henri III and a nun. Whether or not it was true, the scandal was so great that the King “should stop going to that monastery altogether”, and the nuncio is to press this on him hard. The companion letter adds that he is not to raise the matter of the nun with the Queen.

Italian

La voce et l’opinione della pratica del Re Chr.mo con la monaca è uscita tanto publica che se ne ragiona hormai per tutto

The talk of the Most Christian King’s affair with the nun has become so public that it is discussed everywhere.

Como to Dandini 1580

03

17th c.Sloane MS 3188not a cipher

Twenty-five “secret letters” that are notes on John Dee’s conversations with angels

DECODE lists Sloane MS 3188 ff. 109–169 as twenty-five ciphertexts with no decipherment, sender and recipient unknown. They are bound straight after Dee’s own diaries of his spirit “actions”. They are a later reader’s notes on those diaries, in seventeenth-century English stroke shorthand with the names written out in longhand. The leaves hold an index of the Actions, a list of Dee’s writings, and a note on Kelley and Prague dated 1595. There is no cipher, no letter and no correspondent in them.

longhand inside the shorthand

1582 … E.T. … Uriel … Reuchlin · Reconciliatio Kelliana

Shippen’s notes on Dee

04

1702–23Blenheim Papers, Add MS 61575not a cipher

A “numerical cipher” with no numbers, thanking someone for cipher lessons

DECODE catalogues folio 95 as a numerical cipher with no decipherment, by “Frederic, Re di Prussia”. There are no numbers on the page. It is seven lines of plain French, spelled by ear and chopped into syllables, and the writer is thanking someone for showing him how to write in cipher. Judging by the spelling, the lessons went better than the dictation.

French, as spelled

moy en me mon trant d’es cria re

…for the trouble you took with me in showing me how to write in cipher

A lesson in cipher

en me mon trant
d’es cria re
· no digits at all ·

05

after 1688The Wallachian court → a “Most Reverend Father”existing decipherment

A cipher used for nothing but insults

The Szembek papers at Kórnik hold an “encrypted letter” catalogued as an 18th-century letter to the Szembek family. It is the postscript of a letter from the Wallachian court, written after the death of Prince Şerban Cantacuzino in October 1688. The Latin is in clear, and only single words are put into two-digit numbers. A second hand wrote the plain word above every run, and the key rebuilt from those glosses (22 numbers, no conflicts) reproduces them. The words the writer chose to hide are mostly names and insults. The imperial Jesuit envoy Antide Dunod, he says, led the late prince astray, and he gets five enciphered words to himself. The postscript also reports that a certain Gegenbey was killed and his head sent as a gift.

Latin, cipher runs as glossed

impostor et Detractor ille et sine conscientia Burgundus Antidius Dunod

that impostor and detractor, the Burgundian without a conscience, Antide Dunod

The Szembek postscript

06

1758“Cypher from Mr Eichel”, Prussianulls only

A sheet of cipher figures that is nothing but padding

This sheet of figures sits in the Prussian papers at Kew. The key the catalogue proposed does not fit it. The key filed in the same volume does, and it begins by declaring every number that contains a 7, plus four whole ranges, to be a null. On that rule 117 of the 139 groups are nulls, and 21 of the remaining 22 land on cells the key leaves blank. It is a dummy sheet, and there was never a message on it.

the key’s first rule

Tous les nro où je trouve un ou plusieurs 7 sont les nonvaleurs

Every number in which I find one or more 7s is a null.

Eichel 1758

07

1524Rome → Charles Vkey from adjacent plaintext · partial

The oldest long cipher on the list, off by exactly a hundred years

DECODE dated two letters from Rome to “April 1424”. That would have made them the longest ciphertext before 1450, and they were picked as a flagship target on that basis. The letter gives its own date on f. 128v: De Roma xviij de Abril 1524. The catalogue’s “Luis Fernández” is the Duke of Sessa, Charles V’s ambassador. Eight sibling letters carry the court’s own decipherments. The key was rebuilt from them, about 110 code groups and the alphabet, and it deciphers the passage word by word except four code groups and one clause.

Spanish

De Roma xviij de Abril 1524

From Rome, 18 April 1524, and not 1424.

Sessa 1524

1424
↓
1524

08

1710Torcy → the plenipotentiaries at Geertruidenbergnot solved

“Mr Blencowe cannot decypher them”, and neither can we

In the spring of 1710 Louis XIV sent two plenipotentiaries to Geertruidenberg to talk peace with the Dutch. Two of the French letters were intercepted and passed to the English, and they are now in the Blenheim Papers. On the back of Torcy’s letter of 3 April, another hand wrote the verdict of William Blencowe, England’s official decipherer. Three centuries later it still stands. The 690 code groups belong to a two-part code of 800 or more values. None of the keys in the volume fits, nor any of the French codes and tables of those years that were tried. The clear opening, which DECODE’s transcription leaves out, was transcribed here, and its last sentence before the figures is about priests and ministers being caught. The French file copy is in Paris and is not online.

French, the last clear words

Les prestres sont traittez bien plus durement en Angleterre lors qu’ils y sont decouverts

Priests are treated far more harshly in England when they are discovered there

Geertruidenberg 1710, still unread

09

1812St Petersburg → Washingtonkey from adjacent plaintext

John Quincy Adams fumbles his own code: “impassioned” for “impaired”

Despatch No. 88 of June 1812 reports that the Russian Chancellor Rumyantsev had suffered a stroke. Worthington Ford printed nine of its lines in 1914 as “not decyphered”. Adams used his own code, not the one historians assumed. Its key was rebuilt here from the State Department clerk’s interlinear decodes on the same microfilm reel and from Ford’s printed decipherments of other despatches, and it decodes all nine lines. Four groups, whose values are certain, spell im-pas-sion-ed where the sense wants “impaired”, most likely Adams’s own slip at the codebook.

English, syllable by syllable

if his ac-tive in-flu-ence had not been im-pas-ion-ed before it, it can scarcely fa-il to have been af-fect-ed

Erving & Adams

10

1848Rome → Innsbruckciphertext-only break

The Pope’s secret counter-order: intercepted, published with a prize, and a day late anyway

On 15 June 1848 Cardinal Soglia sent 955 digits to the nuncio at Innsbruck, where the Emperor had fled. Mazzini’s L’Italia del Popolo intercepted the despatch, printed it, and offered “a prize of some size for the best interpreter of these figures”. Nobody claimed it. In 2014 it was posted on Cipherbrain as an open problem, and no decipherment was found in print. The key was recovered here from the ciphertext alone. Its code section is alphabetical: Imperatore 8319 comes before istruzione 8340, which comes before Monsignor 8374. The order tells the nuncio to stay with the Emperor wherever he goes. According to the Dizionario Biografico, he had left Innsbruck on 14 June, the day before it was sent.

the reading, in English

…of the Holy Father that you remain at Innsbruck with the Emperor and wherever he may go

Soglia 1848

02 Ten that matter to history

Sieges, assassinations, a bought crown and a king’s change of plan. In three of these the key was rebuilt here from plaintext beside the cipher; the other seven were deciphered with keys that already existed, several of them for passages never printed in clear.

01

1425Urbino → Florenceknown key applied

1425: the oldest cipher deciphered on this site

Early in 1425, Galeotto da Ricasoli wrote from Urbino to Florence’s war magistracy. Gabbrielli glossed a few words and made a key in 1863, which Meister printed in 1902, but the letter itself was never printed and no decipherment of its three long cipher runs was found. Gabbrielli’s key, extended from his glosses on this letter and three siblings, deciphers 97 of the 99 cipher symbols. Madonna is the Countess of Urbino, Caterina Colonna, niece of Pope Martin V.

Italian

perché dice che Madonna è pure femina e sospetosa

because he says that Madonna is, after all, a woman, and suspicious

Ricasoli 1425

02

1527Ghinucci → Cardinal Wolseykey from external plaintext

“To perpetual division”: the English break with Rome, seven years early

Girolamo Ghinucci, Bishop of Worcester, was Henry VIII’s envoy at Charles V’s court. His duplicate of 22 July 1527 is noted in Letters and Papers as having no decipherment, and DECODE has it as not decrypted. The cipher is unusual: a sign stands for a consonant and a small digit above it picks the syllable. Three of Ghinucci’s other letters in the same cipher were deciphered by Wolsey’s clerks, and their decipherments survive in print or between the lines. Lining those up gave the key, and 797 of the letter’s 814 signs (97.9%) now give Latin. The Emperor’s men want to part Henry from Francis I, and the French are said to have offered three millions for the King’s sons. The rumour at court is that Wolsey is going to France to cut the English and French churches off from Rome, permanently. The Act of Supremacy came in 1534.

Latin

ut ecclesia Anglicana et Gallicana a Romana separetur … sed ad perpetuam divisionem

that the English and the French Church be separated from the Roman … but for a perpetual division

Ghinucci 1527, 97.9% deciphered

03

1565Messina → Philip IIkey from adjacent plaintext

The relief of Malta, reported in figures during the Great Siege

García de Toledo, viceroy of Sicily, wrote to Philip II on 16 July 1565 with every sensitive word in two-digit figures. No contemporary decipherment is known (Tomokiyo’s “undeciphered”), and whether anyone else has recovered the key is unknown. It is an easy key: two cribs in the clear Spanish around the figures gave eleven letters, and the figures 12–43 run through the alphabet in order. Most of the work was transcribing the 1,930 groups. The letter tells how the small relief force got into the Borgo, and how La Valette’s fire signals turned back a second run of galleys. It also reports that Sicily could no longer pay for the war.

Spanish

el maestre les hizo las sennales del fuego clarissimas para que no entrassen

the Grand Master made them the fire signals, very clear, so that they would not come in

Toledo 1565

04

1573Warsaw → Pariskey from adjacent plaintext

How France bought the crown of Poland

These are the French envoys’ letters to Charles IX during the election that made Henri of Anjou King of Poland. The result is a corrected key: Tomokiyo’s published table proved shifted, and the key was rebuilt from the court decipherer’s margin notes on these letters, half cut away by the binder, one of them car je n’ay pas cinquante escuz. The corrected key then deciphered three election letters. In them, the Poles can be won by money “as are the Germans, their neighbours”. On 9 May the election is carried against the Emperor, Spain, Muscovy and Sweden.

French

d’Espaigne, du Moscovite et du Roy de Suède, qui tous estoient bandez contre vostre Majesté

of Spain, of the Muscovite and of the King of Sweden, who were all banded against Your Majesty

Lanssac 1573

05

1588–89Blois → Romeknown key applied

The papal legate reports the murder of the Duke of Guise

Cardinal Morosini, legate in France, sent 36 cipher enclosures to Rome, all of them written as one unbroken stream of digits. George Lasry recovered part of the key statistically and identified it as the one Meister printed in 1906. Its homophones are one to three digits long and unseparated, so the key cannot be applied mechanically: the stream has to be divided into units, and several divisions are possible. The division was chosen by scoring candidates with a language model of Italian, and 34 of the 36 are deciphered in full. They cover the Estates of Blois, Saluzzo and Henri III’s march on Paris, and they report the killing of the Guises in the King’s own chamber.

Italian

il Duca di Guisa nela camera di S. M.tà è stato per ordine di S. M.tà amazato a colpi di pugnale

the Duke of Guise has been killed with dagger blows, in His Majesty’s chamber, by His Majesty’s order

Morosini 1588

06

1590Mantua → the duc de Neversknown key applied

“The Pope’s death was procured with poison by the Spaniards”

Vincenzo Gonzaga, Duke of Mantua, wrote to the duc de Nevers on 17 September 1590, three weeks after the death of Sixtus V. Tomokiyo had identified the cipher; its key sheet, filed among Nevers’s own papers, deciphers 95% of the cipher unchanged, and no earlier decipherment was found. Mantua passes on the rumour that the Spaniards poisoned the Pope, and welcomes Henri IV’s expected conversion.

Italian

che la morte del Papa sia stata procurata con veleno da Spagnoli

that the Pope’s death was procured with poison by the Spaniards

Mantua to Nevers 1590

07

1570Tutbury → the Towerknown key applied

Mary Queen of Scots writes to Norfolk in the Tower

Mary wrote this letter weeks after the murder of the Regent Moray, and it is the only one in the series with no contemporary decipherment. Tomokiyo reconstructed the key from the deciphered letters of the series and published a partial decipherment of this one. His key was applied here to the whole letter, all 23 lines; the work was transcription. Mary reports that Elizabeth blames her for Moray’s death. Norfolk is to write only through the Bishop of Ross or Lady Scrope, and never in his own hand.

English, as she spelled it

I am asured that sche … sueves me the veyt of Murray death; but God knoueth

I am assured that she lays on me the blame [Scots wyte] for Moray’s death; but God knows

Mary to Norfolk

08

1691Versailles → Piedmontknown key applied

Louis XIV decides to give up Piedmont, in despatches never printed in clear

The 1819 editor of Catinat’s memoirs printed seven despatches in figures and could not decipher them. Bazeries rebuilt the code in 1893 while hunting the Man in the Iron Mask, printed two in clear and parts of two more, and stopped. His table, applied here unchanged, decodes all seven, 12,362 groups; three had never been printed in clear. They include the King’s twenty-page letter of 14 September: bring the army back over the Alps, hold Carmagnole as long as possible to cover the talks with the Pope, then strip it and burn it.

French

je me suis déterminé à préférer le parti solide à l’honorable

I have resolved to prefer the solid course to the honourable one

Catinat 1691

09

1758–60Lisbon → Romeknown key applied

The nuncio’s secret account of the attempt on the King of Portugal

On the night of 3 September 1758, José I of Portugal was shot at in his carriage. The first of eight ciphers from the Lisbon nuncio Acciaiuoli is his secret account of it. They were deciphered with George Lasry’s 2020 key, which was already on their DECODE records; 25 code groups were added from three leaves deciphered at the time.

Italian

il Teixeira ferito a morte … il Re molto ferito del braccio destro

Teixeira mortally wounded … the King badly wounded in the right arm

Acciaiuoli 1758

10

1807Erving, Madrid → James Madisonkey from external plaintext

Spain’s first minister, cheated by Napoleon, thinks of England

George W. Erving, the American chargé d’affaires at Madrid, wrote this private letter to Secretary of State Madison in the old legation code of Charles Pinckney, and nobody wrote a decode between the lines. Founders Online prints the letter decoded, with ten blank gaps and without the code groups. The 1,145 groups were transcribed from the microfilm and matched against that text, which rebuilt 414 rows of the code; 45 of the 47 groups it shares with an independent 1803 decode of Pinckney’s own despatch agree. What the code hid is Godoy’s fury. He had paid for a bargain over Hanover out of his own funds, and now saw Hanover handed to Westphalia. By March 1808 French troops were in Spain and Godoy had fallen.

English, decoded

believes that he has been deceived and is furious … as a dernier resort he will make an alliance of some sort with England!

Erving 1807, 98.3% of groups

03 Ten that are simply ingenious

Invisible ink, decoy words, a monk’s vowel shift, and the detective work that turns a mislabelled record into a story.

01

1792Berlin → The Hagueknown key applied

A cipher whose only job is to say that the real letter is in lemon juice

The Dutch envoy van Reede wrote from Berlin on 4 February 1792. Deciphered in full with the 1782 Grand Chiffre, the key of his earlier letters to the Prince, the letter is a cover note. The four blank half-pages of its enclosure carry writing in lemon juice, the Prince of Orange is to be told, and “the following figures are nulls”. Those 21 nulls decode to a planted jumble of Conseil d’Etat, la Russie and Archiduc, which is bait for anyone who opens the letter.

French

il y a du citron sur les quatre demi-pages blanches de l’incluse

there is lemon juice on the four blank half-pages of the enclosure

Van Reede 1792

02

1933Shanghaino message

$300,000,000 in gold bars, and almost exactly ten of every letter

Seven gold bars, said to have been issued in Shanghai in 1933 to certify a deposit of over $300,000,000, carry sixteen strings of Latin letters. For ninety years nobody found a message in them. Their letter counts are the problem: against a flat distribution they give a chi-squared of 1.25, where honest random sampling gives about 25. No cipher can produce that. Substitution and transposition keep the skew of the plaintext, and a one-time pad lands near 25. Someone counted out the letters and arranged them to look like cryptography.

the test

χ² = 1.25 on 25 degrees of freedom, where chance predicts 25

Gold bars

03

1635–36Suffolk, a private recordciphertext-only break

A father’s secret log of his baby son’s fits

DECODE has this as a letter of 1635, “unknown sender to unknown recipient”. It is Sir Simonds D’Ewes, the antiquary, writing Latin in a private alphabet he invented at school. The page is a log, one line per convulsion, of his infant son Clopton from 3 April 1635, with the day, the hour and the phase of the moon. No plaintext of the slip exists, and no decipherment was found. The alphabet was recovered from the ciphertext alone, starting from probable words such as convulsio and plena luna, and 96.5% of the words are deciphered. The weekdays and moons all check against the calendar. Clopton died on 9 May 1636.

Latin

in quorum tertia os ipsius pene ad aurem dextram convulsum, quod ante numquam observavi

in the third of which his mouth was drawn almost to his right ear, which I had never seen before

D’Ewes’s cipher log 1635

04

1548Augsburg → Constantinopleknown key applied · partial

Decoy words: nulls that look like Latin

Ferdinand I wrote to Malvezzi, his resident at the Ottoman Porte, in a cipher of graphic signs padded with whole words that look like part of the Latin: etiam, idcirco, atque, Porro. The key was already on DECODE, filed as a separate record at the same shelfmark; applied here, it deciphers the ciphered passages in part. The letter shows the French trying to wreck the Habsburg–Ottoman peace, and Malvezzi’s pay: 4,300 ducats at once.

the decoys

etiam · idcirco · atque · cum · Porro

Ferdinand I to Malvezzi 1548

05

1517Strängnäsknown key applied · partial

A monk signs his chant book in code

The scribe who copied this Dominican sequentiary in 1517 dated it in a colophon where every vowel is written as the letter after it: a becomes b, e becomes f, i becomes k, o becomes p and u becomes x. The Uppsala catalogue printed the rule and the decipherment in 1992; the rule, applied here to a new transcription, gives the same text. The quotation below starts as the scribe wrote it.

Latin, vowels shifted

presens liber sequencionarius est scriptus in conuentu strengenensi

this sequentiary was written in the convent at Strängnäs

Strängnäs sequentiary 1517

06

1616Strassburg, a printed bookalready in print

A “cipher letter” that is a novel, where one of the characters reads the cipher aloud

DECODE R4502, catalogued as a letter with no decipherment, is a copy of the Chymische Hochzeit Christiani Rosencreutz (Strassburg, 1616). Its four “cipher pages” are inscriptions in the story, cut in an invented alphabet. A paragraph later the page boy reads both long ones aloud. The two lines of signs are the dates 1378 and 1459, which Foxcroft’s English edition of 1690 already gives.

German

HIE LIGT BEGRABEN VENUS DIE SCHONE FRAW SO MANCHEN HOHEN MANN UMB GLUCK EHR SEGEN UND WOLFART GEBRACHT HATT

Here lies buried Venus, the fair lady who has cost many a great man his fortune, honour, blessing and welfare

Chymische Hochzeit inscriptions

07

1589The Doge and Senate → Giovanni Mocenigoexternal key matched · partial

A Spanish codebreaker’s lost solution, redone with Venice’s own key

Philip II’s cipher secretary, Luis Valle de la Cerda, deciphered this Venetian letter at Turin in 1589 and kept it in the dossier of his services, now in Madrid. His solution is lost. The key turned up in another country’s archive: the Council of Ten’s Zifra Prima of 1577–78, a sheet in Venice, fits with two added values. Groups such as a51 or c75 stand for letters, syllables, whole words and nulls in one table. The only scan blurs the digits in a regular way, so the transcription was corrected by hand, then by a calibrated decoder and by fitting proposed lines, each step checked against a control. France had asked Venice to help Henry III; the Senate answered that it would plead with Sixtus V instead, to put out the fire of war. 981 of the 1,565 groups (62.7%) are deciphered; the rest wait for a sharper scan.

Venetian Italian

presso lo Sommo Pontifice la intercessione della Republica Nostra … nell’aquetare le sollevationi di quel Regno

our Republic’s intercession with the Supreme Pontiff … to calm the risings in that kingdom

The Ottobon letter, 1589

08

1489–90Buda → Milanknown key applied · partial

A key published only as numbers, turned back into symbols

In 2016 Judit W. Somogyi reconstructed the Milanese ambassador’s key, but printed it only as numbers, without a picture of a single symbol. Her numbers run in order of first appearance in the letter of 2 April 1490, which survives with its clear copy. Matching that letter to its copy gives back every symbol’s shape, and the key can then be applied. It deciphers part of the despatch of 22 November 1489, on peace with Frederick III, the John Corvinus marriage, and Queen Beatrice’s “insatiable wrath of Juno”. The despatch also warns that duplicates of intercepted cipher letters had been uncovered.

Italian

Le littere in cifra che forono intercepte, le quale duplicate sono state scoperte ancora qua

the cipher letters that were intercepted, whose duplicates have been uncovered here too

Buda 1489

09

1586Angoumois → a League leaderexisting decipherment

The “unread” part was deciphered in 1586

Two thirds of this anonymous League letter of October 1586 were listed as having no decipherment. On the next leaf, f. 168, is a clear text of the cipher down to a cross mark. After the cross, the decipherer had switched to writing the plaintext between the lines of cipher. The glosses confirm Tomokiyo’s key (87 = e). The letter says that Fumé, vice-admiral of Guyenne, keeps proposing a reconciliation between Navarre and the League leader it is written to.

French

Fumé, visadmiral de Guienne, tres confident du Roy de Navarre, bien qu’il soit catholique

Fumé, vice-admiral of Guyenne, very much in the King of Navarre’s confidence, although he is a Catholic

Clairambault 357, 1586

10

1824Snell → an unnamed addresseekey from adjacent plaintext

Questions in invisible ink, and the official who made a hundred attempts

Snell wrote this message in sympathetic ink between the lines of a harmless letter. It reached the Hessian authorities, and an official copied the six lines of cipher onto a leaf now in Marburg, which HCPortal lists as a polyalphabetic cipher with no decipherment. Under the copy the official added a note. He had the key table beside him but could not use it, until “after a hundred attempts” he saw that the letters b c d e f g along its top are the key, written over the cipher again and again. The rest was worked out here: a Vigenère table without j whose alphabets run on past z into the digits 1 to 4, and the copyist’s slips. All 169 signs read. The message asks for Lisching or someone else to be sent, whether “the discovery of this plan in Wetzlar” had consequences, and what news there is of America.

German

Was hat Paul Follen wegen America von Fries, Oken, de Wette gehört?

What has Paul Follen heard from Fries, Oken and de Wette about America?

Snell 1824, all 169 signs

04 Six of the hardest breaks

No key, no crib and no earlier decipherment: each of these keys was recovered from the ciphertext alone. They were chosen for difficulty, not for the story; how they were picked is below. Two more breaks of this kind are already among the thirty: Soglia 1848 and D’Ewes 1635.

01

1637Wesel → Landgrave Wilhelm V of Hesse-Kasselciphertext-only break · partial

Ten letters from a war in Westphalia, with no key among 176 key sheets

Otto von der Malsburg, Hesse-Kassel’s general war commissary, reported to his landgrave in 1637, the year the landgrave died in exile. The archive’s portal lists all ten letters as not solved. None of the 176 Hesse-Kassel key sheets fits them. The cipher is German in two-digit groups, with two to seven groups for each letter, capital signs for pairs such as ch, st and sch, and no word division. Annealing one page alone failed. Three letters pooled, 1,575 groups, gave the same key from every start. About 9,700 groups were transcribed from 18 scans, and 236 doubtful ones were checked again on the images. The letters cover garrisons without pay, the siege of Ehrenbreitstein and the refusal of safe-conducts to the Cologne peace talks. No reading was found anywhere before this one.

German

das ich auch noch nit weiss, wo hervor unsere garnisonen hiernechst der unterhalt genommen werden soll

that I still do not know where the upkeep of our garrisons is to come from now on

Malsburg 1637, 95.9% of 9,736 groups

02

1560the Agenais → the connétable de Montmorencyciphertext-only break

Ten pages of an invented cursive script, and the clue in a pair of strokes

The only clear words on these ten pages are the date, 12 December 1560. Everything else is a cursive pseudo-script, 1,715 words with no gloss, no key and no edition. The first solver, one sign to one letter, produced gibberish. The way in was a statistic: one sign was followed by the same second sign in 262 of 283 cases, so the pair had to be a single unit. Fitting the commonest word to que broke it open. A re-check against the images then found that about 300 signs first read as one letter were a smaller look-alike for another. The mémoire describes the Protestant rising in Guyenne in the first days of Catherine de’ Medici’s regency. George Lasry wrote that he had broken it independently in 2023; his solution has not been published.

French

Despuis les derniers advertissemens que envoiay a vostre grandeur, les affaires de pardeça ont couru de mal en pis

Since the last reports I sent to your greatness, affairs here have gone from bad to worse

Mémoire 1560, 98.9% of the words

03

1585–86Henry of Navarre → Ségurciphertext-only break

Henry of Navarre raises a German army, in a table found by counting in fives

Three letters in figures to Ségur, the envoy raising a German army, were listed as letters of Henry III. Their 461 figures were transcribed from the Gallica images. Taken modulo 5, the higher numbers all fall in one slot of every five. That is the trace of an alphabetical syllable table, ba be bi bo bu, ca ce ci co cu. A solver built on that structure, with the letters’ clear French as context, converged on a key whose letter part came out in alphabetical order by itself. It scored 97% on a matched control. The letters are signed Henry at Montauban and are Henry of Navarre’s. Lasry wrote that he broke them independently in 2024; his solution is unpublished.

French

capituler et conclure promptement avec ceux que vous adviserez, s’il est possible avec le duc Casimir

come to terms and conclude promptly with whomever you think best, with Duke Casimir if possible

Navarre to Ségur 1585–86

04

1754–56The Hague → Dresdenciphertext-only break

The intercepts the Dutch cabinet could not read, because the figures never stop

The Dutch post office opened the Saxon resident’s letters and copied them for the Stadtholder. The cabinet gave up on the figures because the groups “are not separated, by blanks or dots as is usually the case.” The first problem was where one group ends and the next begins. The digits 5 and 8 never occur inside a group, so they are nulls. Without them the stream falls into about sixty two-figure codes. A homophonic solver found nothing. The way in was Kauderbach’s own key of 1761, kept in Dresden. Its numbers are different, yet its list of letters and syllables fits, and an assignment search onto that list converged twice to the same key. In each of the ten letters, 65 to 97% of the cipher decrypts to sense, including the news of the Treaty of Versailles of May 1756.

French

une consternation extraordinaire ici

an extraordinary consternation here

Kauderbach 1754–56

05

1619Bonn → Munichciphertext-only break

Ten pages of figures with no sender, no recipient and no language, and the letter names itself

DECODE knew only the year of this record: no author, no recipient, no place, no language. The cipher is a homophonic substitution on the even numbers 10 to 78, with two word-end signs and a three-digit nomenclator. Its key was recovered by annealing 2,482 letters against a 16th-century German model, with about sixty single-digit transcription slips repaired. The same key reads the sibling letter in the file unchanged. The decipherment then names the writer: Ferdinand of Bavaria, Elector of Cologne, to his brother Duke Maximilian, six weeks after the Winter King’s coronation, on quartering and paying the Catholic League’s troops. Neither letter was found in print.

German

sich vor des andern Theils Einfal, Gwalt und Betrangnus zu entweren

to defend themselves against the other side’s incursion, violence and oppression

Ferdinand of Bavaria 1619, 95.8% of the tokens

even figures 10–78
“in meiner Statt Bonn”
· signed Ferdinand ·

06

after 1580“L’Estat du Roy de Navarre”ciphertext-only break

The letter m was a cluster of dots, and the fault was in the solver

DECODE catalogued this memoir as “unknown sender”, perhaps in English. It is French, about fifty graphic signs with three to six homophones for the common letters. Every earlier pass took one sign for punctuation: the letter m is written as a group of two, three or four dots. Even after that the key would not come. A control experiment, real French enciphered under a key of the same shape, located the fault: the solver, not the manuscript. With one wrong constraint removed and its temperature fixed, the control solved completely, and the real text broke at once. It reads at 95.1%, on Navarre’s party, Parma and the reiters.

French

un party de malcontens

a party of malcontents

Harley 1582, R8505

05 How these were chosen

The thirty come from the three hundred or so write-ups on this site and the notes behind them. They were chosen for the story, and difficulty played no part; the full list sorts by outcome, and the hardest breaks are ranked in the repository’s solved ranking. Each badge names how the text was obtained, in the terms of the site’s glossary. Ciphertext-only break (two of the thirty: Soglia 1848 and D’Ewes 1635), key from adjacent plaintext (six) and key from external plaintext (two: Ghinucci 1527 and Erving 1807, matched against a decipherment or an edition printed elsewhere) mean the key was rebuilt here. Known key applied (ten), external key matched (one) and existing decipherment (three) mean that a key or a decipherment already existed, and the work was transcription and decipherment. Partial means some of the text is still open. Quotations keep the original spelling, and the translations follow the write-ups. Four of the thirty are not ciphers at all, or hold no message; one was deciphered in print long ago, and one, the Geertruidenberg letter of 1710, is still unread. The selection was redone on 25 October 2026 from the larger list: six cards gave way to finds made since 2 October.

The six hardest breaks were chosen the other way round, by difficulty alone, from the targets whose key was recovered from the ciphertext alone. The tests: no key, crib or decipherment used; no sibling key; an unknown system; the cipher transcribed from the images here; the reading measured (95% or more of the cipher, except Kauderbach, at 65–97% per letter); and a control, a sibling letter or a re-check on the scans behind the key. Breaks first made by someone else and checked here, such as the Mercy instruction of 1648 (NoAutopilot), are left out, as are four that turned out to be independent re-solutions of a cipher someone had already broken. Two of the six, the Mémoire of 1560 and the Ségur letters, had been solved privately by George Lasry, who never published them; the cards say so.

Every image is credited on its card to the archive or edition that holds it. Select an image to open it full size.

06 Sources

• The write-ups linked from each card, with their working notes in the repository, one folder per target.
• The DECODE database (Megyesi et al.), where most of the records live, and the published keys of George Lasry, Satoshi Tomokiyo, Judit W. Somogyi and Aloys Meister.
• The printed editions named on each write-up, among them Rommel 1840/1846, the Mémoires de Catinat (1819) and Barwick’s Life (1721/1724).