Venetian ambassador at Rome to the Doge, 29 April 1628

At a glanceKey from external plaintext · complete

The letter: A Venetian ambassador’s despatch from Rome to the Doge, 29 April 1628 (“Ser.mo Principe”); DECODE gave no sender, recipient or transcription.
Where: ASVe Busta 27 f. 354; DECODE R1862, four images.
The cipher: A syllabary grid: the prefix gives the vowel, the last two digits the consonant, with single letters, a small word list and paragraph marks.
How it was read: The same cipher appears in R1874 (Mocenigo, Madrid, 1628), imaged with the chancery’s decipherment. Aligning that plaintext to its cipher by expectation-maximisation gave the key, then applied here with word codes added from context.
What it says: Cardinal Barberini twice asked whether Palmanova was well fortified and the Pontebba pass secured. The rest covers Wallenstein, the aulic council, and pressing the Pope to go as legate.
Still open: 99.1% of the 1,280 code groups have a value and 0.95 of the text gives sense. Twelve word codes and one line stay open.

01 The record, and the sibling that carried the key

DECODE R1862 is four images of a despatch at ASVe, Busta 27 f. 354, dated 29 April 1628 and placed at Rome. The record carries no transcription, no sender and no receiver, and the archive’s images are not in the public domain, so nothing of them is reproduced here. Cipher and clear text alternate on the page: page 1 and page 2 are almost entirely in figures, page 3 runs clear for eleven lines and then ciphers again, page 4 opens and closes in the clear with two ciphered blocks between.

The way in was a sibling. DECODE R1874 (ASVe Busta 30 f. 174) is a despatch of Alvise Mocenigo from Madrid, 26 October 1628, in the same cipher; its sixteen images hold the ciphered despatch, a second enciphered copy of it, and, on images 8955–8959, the chancery’s own decipherment, paragraph by numbered paragraph. An earlier attempt on R1874 had transcribed that decipherment but had not used it to recover the key.

02 Rebuilding the key by alignment

The decipherment of R1874’s paragraphs 5 to 16 was transcribed here, about 4,800 letters of plaintext. DECODE’s transcription of R1874’s cipher was then split into five stretches matching those paragraphs (the ratio is a steady 1.84 letters per group), and each stretch was aligned to its plaintext by EM over the possible divisions of the plaintext into group values, sharpened each round until every group settled on one value. The solver was checked first on a synthetic syllabic cipher built over the same text, which it recovered exactly (97 of 97 groups). The first run on the real data collapsed, because a group was allowed to stand for nothing and the frequent groups went null; capping values at four letters and forbidding nulls fixed it.

One detail decided the shape of the answer. DECODE’s transcriber wrote the figure 5 in two forms, as “13” and as “21”. Folding both to a single 5 gave a muddled key; keeping them apart made the table regular at once, because they are two different rows of it (51 and 53 in R1862’s clearer hand).

03 The key

The cipher is a syllabary laid out as a grid. The prefix gives the vowel and the last two digits the consonant:

vowels: 6 = a, 51 = e, 62 = i, 53 = o, 64 = u (61 is a second i/e row)
consonants: 07 b, 08 c, 09 l, 10 d, 11 f, 13 g, 14 gn, 17 gr, 19 l, 20 m, 21 n, 22 p, 23 pr, 24 qu, 26 ss, 27 r, 28 s, 31 st, 33 t, 34 tr, 37 v, 38 z; so 6210 di, 5133 te, 5333 to, 6433 tu, 5308 co, 6220 mi
single letters: 680 a, 6280 e, 6480 i, 5190 o, 6490 u, 692 n, 6292 r, 6482 l, 5390 s
word codes: 670 che, 672 con, 6270 in, 6249 il, 5102 del, 5302 nel, 5124 and 5387 quel, 5384 qual(che), 5374 per, 5142 duca, 6440 sua, 6442 suo, 6240 ave, 5373 part, 6477 poc, 6287 non, 5347 papa, 6247 Cesare
paragraph marks: 698, 5198, 5199 stand at the edges of the ciphered passages, as 6499 does in R1874

The word band is alphabetical. Above column 39 the codes of each prefix row run in alphabetical blocks — row 53 gives papa 47, part 73, per 74, piu 77, qual 84, quel 87, ques 88; row 64 gives signor 39, sua 40, suo 42. That brackets every code that is still open, and it is how several were identified: 6274 falls just after in and is “molto” (“star molto oculati et avvertiti”), 5140 falls just below duca and is “dov” (“dovrian star”). It also caught three values inherited from the R1874 alignment that were wrong: 628 is sa, not “n” (“non sarian giamai”, “fatto sapere”); 6439 is signor, not “mini” (“Sua Signoria Illustrissima”, “il Signor Cardinal Barberino”); 6481 is a second i.

The grid cells that R1874’s decipherment never reached were filled by the structure itself (a cell’s value is its column’s consonant plus its row’s vowel), and the word codes were identified from context in this letter: 5384 by “cavar anche qualche intimo particolare”, 5347 by “a fin che il Papa deliberasse”, 6240 by “avendomi ricercato”.

04 What the despatch says

It opens in the clear and turns to cipher in the middle of the first line:

Opportunissima occ.ne ho io havuta hieri col Cardinal Barberino di parlargli di quel negotio che lei mi si è commesso con cotesto Eccelso Conseglio, et di cavar anche qualche intimo particolare desiderato da lei…

What the ambassador had been asked to find out, and what the Cardinal had asked him, are the heart of it. Twice the Cardinal had wanted to know

…perché due volte nelle capele [papali] e avendomi ricercato con [più] instanza se Palma sarà ben fortificata, quanto presidio aveva, se il passo della Pontebba si era ben assicurato…

(“…because twice in the papal chapels, and having asked me with [more] insistence whether Palma will be well fortified, how large a garrison it had, whether the pass of the Pontebba had been well secured…”) Two words in it were supplied on 30 September 2026, both grade I. The code 5349 has no value from R1874, but the word band is alphabetical and 5349 stands just after 5347 papa and before 5373 part: after nelle capele that gives papali, the papal chapels where the ambassadors met the Cardinal (or the stem papal, with the next group’s e as its ending). Before instanza the clerk wrote 6202, which is bi (as in du·bi·tando on page 2) and gives no word; read as one of his one-digit slips for 6222, pi, it gives con pi[ù] instanza. A slip for an unknown word code 6242 (between ave and Cesare, so a b- or ca- word such as calda) cannot be excluded.

That is, whether the Republic’s great fortress of Palmanova and the Alpine pass above Venice were ready. Hearing from other quarters that “Imperiali uniti con Spagna parlavano con molto sprezzo” of the Republic, the ambassador answered by asking, “con intiera confidenza”, whether the questions had been put “a qualche fine”, and begged that anything of the kind be told him, so that the Serenissima might know in time “perché da lei fossero trovati i rimedii oportuni”, but without naming “quelli che trattassero simili” matters.

The rest of page 2 is the substance of the audience, and it turns on secrecy: such dealings “non sarian giamai […] palesi, ma ben sotto profondissima segretezza”, and the Cardinal spoke “aprendosi in strettissima confidenza”. He had “havuto particolar notitia de pensieri vasti”; he had heard of the disgusto di Aquileia, the Venetian–Imperial quarrel over the Patriarchate; and of arms and men “che si trovavano in piedi nella Germania”, which left him “dubitando avessero qualche pensiero di molestare la Republica dalla parte del Friuli in particolare”; and he told the Cardinal that “quelli Signori dovrian star molto oculati et avvertiti”, so that he “non potei far di meno di non dimandarle” more. The Cardinal “interrogavami per dir il vero”, and the ambassador “mi sono andato certificando” that even if Spaniards and Imperialists had set “l’animo contro la Republica”, they “non sono in tu[tto]… effettuar i loro… disegni”.

The clear paragraph at the foot of page 2 and the top of page 3 carries the Cardinal’s argument that Venice need not fear an imperial descent: the Emperor ha pur troppo che fare in casa propria, having to watch the King of Denmark, Bethlen Gabor, the Dutch and perhaps the Turk; Wallenstein would not come into Italy without his whole army, and quartering it would take gran summa d’oro, di che l’Imp.re n’ha gran carestia. The ciphered part of page 3 reports the Cardinal on the imperial aulic council, “nel qual, se ben la maggior parte de’ consiglieri sia aderente al Re di Spagna”, and that he shows himself “più tosto amorevole che difidente”. It ends with the Cardinal offering “un particolare da esserne gran ri[f]lesso” and asking, in the line that had been missed out of the transcription altogether, “la prego non nominarme”: do not name me as the source.

Page 4, in the clear, lays the attempt on Genoa at the Duke of Savoy’s door, and argues that Spaniards and Imperialists, ill satisfied with him, will trust him less. Then, in cipher, the ambassador’s own move:

…ringratiai il cardinale di co[sì] confidenti communicatione, et sopra […] l’ultimo del Duca di Sivoia presi ocasione di caricar la mano a fin che il Papa deliberasse mandar la sua persona legato […] per acomodar gli interessi di Mantova e di Savoia insieme…

He pressed “la facilità che già si averia trovata nel Duca di Savoia di ridursi”, and the glory for His Holiness of “questa opportunità di dar pace”. The last ciphered lines return to the Cardinal, who “have dato il sudetto aviso… di Genova”. All of this is late April 1628: Vincenzo II Gonzaga was dead, Nevers was in Mantua, Spain and Savoy had agreed to partition Monferrato, and Urban VIII’s nephew was running a papal policy of professed neutrality.

05 How the transcription was made, and what is still open

The record has no transcription, so the 1,274 code groups were transcribed from the images here and then checked, line by line, with the key in hand: a line that does not decipher into Italian is a line to look at again. Five such passes were run at first. Two traps in this hand mattered: a final n-shape is a 2, not a 1, and a 7 is a “>” with a long open tail while an 8 is a hook plus a closed loop. Page 2’s lines bow by up to 35 px, and two of its transcribed rows turned out to have been blended from the head of one physical line and the tail of another, with a third row belonging to no line at all; tracking the baselines and re-splitting fixed the alignment, and the join it produced across two lines, “andavo dubitando avessero qualche pi | nsiero”, confirms it.

Eleven passes of that kind were run in all, and two of them changed the shape of the text: one manuscript line on page 3 had never been transcribed at all (it reads “la prego non nominarme” — I beg you not to name me), and the lines slope enough that a fixed horizontal band picks up the neighbour’s ink on the right-hand side, which is what produced the blends.

The clerk miscodes as well as misspells, always by one digit, and reading him means reading through both. His spelling slips stand in the plain text — sivoia for Savoia, germinia for Germania, dovore for dovere, partcolar, soto, conunicarla. His encipherment slips are recorded separately (the transcription keeps the digits he wrote, and the key file lists the group he meant): 6220 for 6223 in esprimersi, 632 for 672 in confidenza, 5337 for 5137 in altrove, 5388 for 5308 in così, 68 for 637 in Mantova, 537 for 627 in considerati, 6480 for 6490 in più, 6222 for 622 in preparati.

What remains: twelve group tokens (0.9%) have no value: 6443, 6279 and 6478 twice each, and 5349, 5370, 51708, 5378, 5379, 6241 once each (5349 now has the conjectural value papali from its alphabetical place, section 04). All of them sit in the word and name band, the part R1874’s decipherment does not cover, so only another letter in this cipher would settle them: none is digitised on DECODE, and the Venetian key registers that are (the Codice Amadi, ASVe Inquisitori di Stato reg. 1269, and the Busta 4 registers) are all of the previous century. One line, P2 L22, is image-true after four independent readings and still gives no sense; every group in it is a structural grid value, so it turns on one of the clerk’s slips or on a word outside the corpus. The ambassador himself is still anonymous: he does not name himself, and the Calendar of State Papers Venetian, which calendars only the England-related despatches, has nothing from Rome for 29 April 1628.

The extent was measured. Each line’s decipherment is scored against an Italian character model of the period, calibrated on genuine Italian of the same length taken from R1874’s decipherment (real text of this length falls below −2.25 per character only 5% of the time), and scored with the edges of its neighbours so that a word split across a line end is not counted against it. By that measure 0.95 of the code groups stand in lines that give sense; a stricter 25-character sliding window gives 0.92 and a hand audit token by token gives 0.96, the spread being the clerk’s own spelling. That meets this project’s threshold for a complete decipherment (95% of tokens giving sense, no gap untried, only scattered codes open).

Full texts

The transcription and the decryption of every document on this page, so that the reading can be used and checked. Short files are shown here; all are in the repository.