I noticed there has been active scholarly research recently about ciphers related to Emperor Charles V (and Philip II), of which I was not aware in writing "Ciphers during the reign of Emperor Charles V" (2011-2025).
Kolosova (2017): Charles V (1521-1527)
Luo (2021): Isabella of Portugal and Others (1530-1539)
Bertomeu Masià (2025): Imperial Interests in Florence
Cascella Alcaraz (2025): Marquis of Aguilar
Reales Pérez (2025): Mary of Austria (1546)
Pich-Ponce (2023): Simon Renard, Granvelle, and Charles V
Pich-Ponce (2025): Simon Renard, Philip II, and Emmanuel Philibert of Savoy
Luo (2025): Joanna of Austria (1554-1559)
Florio (2025)/Benavent (2025): Margaret of Parma (1566)
|
Olga Kolosova (2017), "El lenguaje secreto de la diplomacia de Carlos V (1521-1527)", dissertation, Universitat de València, 854 pp. Olga Kolosova (2024), "El lenguaje cifrado en tiempos de Carlos V (1521-1527)", Ediciones Universidad de Salamanca, DOI: https://doi.org/10.14201/0MX001 234 pp. |
Kolosova reconstructed 17 ciphers used in letters to Emperor Charles V in 1521-1527. (Cifra 1 to Cifra 17 in the thesis are designated Ko.1 - Ko.17 herein.) Kolosova (2017) discusses each cipher in great detail, which, however, resulted in many repetitions and scattered descriptions. Kolosova (2024) (I've only seen a preview on Google; it appears the purchase page doesn't work in Japan) seems to be organized into chapters dedicated for discussion of specific topics and presentation of the 17 ciphers. From the table of contents alone, it can be seen that updates have been made (e.g., terms used by cryptographers such as "nulls" and "homophones" are introduced).
The following is based on Kolosova (2017).
Guided by the work of Galende Díaz, she chose 77 letters for study from hundreds in Colecció Salazar y Castro in the Library of Royal Academy of History (BRAH), supplemented by letters in the National Library of Spain (BNE). To these is added the text of a treaty between the Emperor and the Pope, making the total 78 (p.6-7).
Limiting the sources to letters to Charles V (thus excluding those between ambassadors) is because the aim was to study ciphers created by the chancery of the Emperor (p.5-6, 96).
Evidence of further five ciphers was found but those were not included in the thesis because of scarcity of data and also because the ciphers were not with the Emperor (p.96).
Charles V's ciphers in this period feature a substitution alphabet for letters and a nomenclature for representing words or word elements with three- or two-letter codes generally arranged alphabetically. These are common in Ko.1 to Ko.11. The similarities testify that Ko.1-Ko.11 were all created by the same person or people, specifically the chancery of Charles V (p.482).
Kolosova's study is from the perspective of philology (p.3, 38) and has her own terminology.
She speaks of four levels of coding: alphabetical (for letters), syllabic (for syllables, developed only in Ko.17), morphological, and lexical (for words) (p.108, 157). The morphological level covers morphemes (an inflectional or derivational component of a word (p.125, 350, 358)) and lexemes (a component of a word with a semantic value, e.g., "negoci-", "guard-", "perde-" (p.125-126)). Of course, boundaries of morphological units (morphemes/lexemes) may not be the same as used by modern linguists (p.126, 237). She also reminds us that "syllables" as used in the study refers to sílabas abiertas (open syllables), which is different from syllables in linguistics (p.9, 481). (To give an example in English, "strength" is one syllable in linguistics but when "re" is enciphered as a unit, it is often called a syllable in cryptology. Later Spanish ciphers often provide for closed syllables like "ren", too.)
An "alphabetical cipher" (cifra alfábetica) is a cipher (mainly) consisting of only a (homophonic) substitution alphabet (p.431 etc., referring to Ko.12-16) (on p.831-835, the term seems to be used as something opposed to a "mixed cipher" below). She also speaks of a grammatical cipher (p.138, 316, 424, etc.), used in similar context but this term seems to put emphasis on the formative nature of the nomenclature.
She appears (I must admit I rely on machine translation) to consider la cifra híbrida gramaticalizada (grammaticalized hybrid cipher), the feature shared by Ko.1-Ko.11, is an important invention of the imperial chancery (p.97-99). By "grammatical", she appears to mean elements (lexemes/morphemes) in the nomenclature are combined to form various inflectional/derivational forms (p.167; see also p.11, 97, 146, 157, 221, 424, 485, 499, 503). To verify the novelty of Charles V ciphers, we need analysis of Spanish ciphers for the Catholic Monarchs (see another article).
A "morphological cipher" appears to refer to a similar but different idea (p.236-239). In my understanding, indication of inflection/derivation by adding a (two- or three-letter) code (a common lexeme + a formative morpheme) is called morphological and indication of inflection by a slightly altered code (e.g., three-letter code with a different vowel in the middle) is called grammatical (also called "synthetic means of conjugation" (p.238, 371-372)). So, while "grammatical" may refer to presence of formative elements in the nomenclature, its combination with the substitution alphabet seems to be called "grammatical hybrid" or "mixed grammatical" or simply "grammatical" (p.99, 386).
Kolosova discusses in great detail what cryptographers may take for granted or just accept as it is. For example:
- Frequent letters have more homophones (p.484-492) (called variantes in the thesis) (I can understand philologists did not like to use the term homophone when the sound is not involved. The term homophone is introduced in Kolosova (2024).)
- Cipher symbols are alphabetical letters, Arabic figures, or "steganographic" symbols (p.119)(I tend to simply speak of arbitrary symbols or graphic symbols even when letters or figures are included, but I'm aware these may not be the best) . A cipher which employs two or more from alphabetical letters, Arabic figures, and steganographic symbols is called a "mixed cipher" [cifra mixta] (p.10, 115).
- Sometimes, I'm not sure of her analysis. For example, rather than minute differences in shapes for symbols representing D, E, L (p.116), I think it more important that the symbol shapes discussed are meant to be "g", "9", "q". (I heartily agree that the similarity in the glyph of these symbols can lead to errors! Although she mentions it has an advantage of confusing codebreakers (p.358), I don't think cryptographical strength should rely on error-prone choice of glyphs.)
- She examines in detail the symbol shapes: one symbol is like a rotation of another (p.165, 290, 197-198) etc.
- She makes much of rasgos de serie (serial features), by which she means arrangement of a group of elements in consecutive three- or two-letter codes in the nomenclature, meticulously discussing presence or absence thereof for various groups (p.128, 168, 176, 323, 371, etc. etc.). She considers absence of serial features or presence of fake series can make codebreaking more difficult (p.131, 169).
Nulls as called in cryptography are called marcadores de la cifra (markers of the cipher) (specifically, marcadores iniciales, intermedios y finales) because they are considered to be used to indicate where ciphertext starts and ends and word division in the ciphertext (p.9,p.107, p.252, p.480, etc.). According to this understanding, when the whole text is enciphered, markers can be omitted (p.461).
This is not how these symbols are viewed by cryptographers. More proper appellation is on p.350, where they are simply called "introductory elements" and "final elements" (not marcadores), with an explanation that it obscures (rather than shows!) where the real ciphertext begins or ends.
Kolosova (2024) introduces the term "null".
One practical question cryptographers face in reconstructing a cipher is distinction of I/J/Y, U/V/B, and S/Z/C.
At the time, virtually no cipher distinguished between I and J, so the real issue is distinction between I and Y. In letters in cipher from this period, we often encounter a symbol used for both I and Y. Similar ambiguity is seen between S and Z (and Ç in Spanish). And X can also be confused with J.
Of course, whether such distinctions are made or not affects the discussion of the number of homophones for each letter (as recognized by Kolosova).
Today, B and V in Spanish have the same sound. In the sixteenth century, the distinction of the sounds B and V was being lost (Language Answers).
Kolosova finds B separate from U/V in 12 of the 16 ciphers (p.495). To me, it was a bit surprising that four ciphers do not distinguish B and V.
In the use of Ko.1, confusion between U/V and B was seen about 10% of the cases (p.109-110). So Kolosova's reconstruction does not have a separate entry for B and gives three symbols under B/U/V.
An interesting case is a comparison between Ko.9 (Alonso Sanchez' cipher) and Ko.7 (Lope Hurtado's second cipher) because the substitution alphabet of the latter is almost a copy of the former. Kolosova considers that one important difference between the two is that B and U/V are distinguished in Ko.9 but not in Ko.10 (p.389).
I think of a possibility that the original cipher had a distinction but the user's intended spelling varied between B and V. For example, for Ko.9, the case may be that Lope Hurtado (or his secretary) had the same substitution symbols as Ko.7 but tended to mix up B and V. Kolosova is also aware that reconstructed ciphers may look different because of "individual preferences of the secretaries" (p.311, 352; see also Bertomeu Masià (2025), n.17 and corresponding text).
Ko.17 (Antonio de Leyva's cipher) is a cipher that both Kolosova and I reconstructed independently. One significant difference is that she considers that there is no distinction of B and V in syllables like BA and VA (p.475), while my reconstruction gave them separate entries. I need to study this further.
For ciphers for which B and V tend to be mixed up, it would be interesting to see if there are any specific words for which confusion tends to occur.
As for I and J, these are treated as the same letter in 13 ciphers (which is consistent with the majority of ciphers at the time), with two ciphers distinguishing I and J and one having incomplete distinction (p.495).
For Ko.2, Kolosova assigns a symbol for J as distinct from I/Y, calling it an "extremely infrequent" distinction (p.158). At the time, the sound /x/ represented by J in modern Spanish was written "x" and pronounced similar to J at the time (Linguism). So "the cipher secretary who composes writes x and the secretary that decodes writes j." (p.159). Then, I think the symbol should better be assigned to X to be consistent of the major practice, even if X is not on record in the studied letters. That is, I think we should have X instead of J in Ko.2.
In Ko.11 (the Adornos' cipher), J is given an independent symbol not only separate from I/Y but also from X (p.410).
As an aside, an interesting observation in Ko.1 is that the homophone used for letter Y and the code used for the conjunction y ("and" in English) are distinguished 95% of the case, a phenomenon unique to Ko.1. The symbl for the conjunction is sometimes used for the letter, but the reverse "contagion" never occurs (p.112).
Another issue about Spanish ciphers is whether to include Ñ and LL as an independent letter in the alphabet. (Today, Ñ is an independent letter in the alphabet.)
According to Kolosova, Ko.1 is the only instance among Charles V's ciphers that has Ñ in the alphabet.
In Spanish, LL was used as an independent letter from 1803 to 2010 (Wiktionary). Further, RR and SS had different sounds from the single letters (Spanish Language, "The Development of Spainsh Sibilants" (pdf)).
For Ko.1, Kolosova finds LL and RR in the alphabet (p.110) but not SS, with words with SS being enciphered with S (p.113).
If she finds some double letters are given independent symbols, I don't dispute it. But there is another question of whether they were part of the alphabet or placed in a dedicated section for double letters.
Ciphers at the time often had symbols for double letters but usually they were separate from the alphabet. In order to see whether LL (or RR etc.) was really in the alphabet, we need to consult original contemporary ciphers. From the latter half of the century, the ciphers of the Viceroy of Peru (1575) and the Governor of Cuba (1591) had LL and RR in the alphabet (see another article). There are none among the 46 ciphers from the time of Philip II printed in Devos (1950) that have LL, Ñ, or the like in the alphabet. For example, Cg.5 provides for double letters "bb cc dd ff gg ll mm nn ó ñ pp rr ss tt uu" after the alphabet and syllables. (If symbols for MM, TT, etc. are found in reconstructing a cipher, it may be more likely that double letters are given symbols separate from the alphabet.)
In the syllabic cipher of Antonio de Leyva (Ko.17), she observes that llguen (arrive) is enciphered as l-le-gu-e-n (p.476).
One topic of interest is how to represent various inflectional forms of verbs. The study of this topic should be best conducted with some original cipher table, because any study based on ciphertext is restricted by actual use of codes. Among the cases studied by Kolosova, Alonso Sanchez' code (Ko.9) gives the best case because of the amount of ciphertext available (p.373; six from the many letters extant were studied (p.350)).
Obviously, many grammatical forms are created by some code (or several code groups) for some base form plus some additional code(s). This is the case in French ciphers I know. Various endings are enciphered letter by letter, occasionally with use of a code for a frequent ending ("-oit", "-ent", etc.). Sometimes, the base form is loosely used for several forms. In later centuries, French codes often explicitly allowed for covering multiple endings (see (8) in another article; Pierrot et al. (2025) quoted in my blogpost; and a Napoleonic specimen in another article).
In the case of Alonso Sanchez's code, Kolosova observes the third-person singular of the present indicative is predominantly used for the base form (p.373). An obvious choice, the root of verbs ("according to the perception of Renaissance secretaries"), comes second (p.379).
While Kolosova seems to have sought a systematic way that inflectional forms are derived from code elements, her conclusion seems correct in that no systematic way was intended by the cipher secretaries, who simply assigned codes according to what they thought practical and efficient (p.380).
The following describes 17 ciphers used in letters to Charles V in 1521-1527 reconstructed by Kolosova (2017).
Ko.1-Ko.11 are mainstream ciphers with a substitution alphabet and a full nomenclature in three- or two-letter codes.
(p.156, Substitution alphabet: p.158, Nomenclature: p.185, Specimen: p.162)
Juan Manuel (Wikipedia; see also bio on p.156-157 of the thesis) was ambassador in Rome. The 77 letters studied include four letters of Juan Manuels from February to March 1521 (no.1-4).
(p.106, Substitution alphabet: p.110, Nomenclature: p.150, Specimen: p.111)
This was used by:
the Duke of Sessa (Luis Fernández de Córdoba (Wikipedia)), ambassador in Rome from September 1522 after Juan Manuel;
Juan Pérez, a secretary who reported illness of the Duke in August 1526 and thereafter the Emperor's chief informant in Rome (p.107); and
Commander Miguel Herrera, mayor of Pamplona (1525-1526) and one of Charles V's informants (p.826).
This cipher was also used in the text of the treaty of 1525 (in Italian) between Pope Clement VII and the Duke of Sessa for the Emperor. (Of course, this does not mean the cipher was shared with the Pope. I suppose it was the treaty text reported (probably in Herrerra's letter no.63) to the Emperor that was enciphered. It would be interesting to see how the code for Spanish was used to encode Italian text.)
As usual, this consists of a substitution alphabet and a nomenclature. (By the way, Kolosova observes that one of the three symbols is rarely used and mentions a possibility that the secretary used the symbol by mistake by the influence of some other cipher (p.110). This is a phenomenon we call "a cross-cipher error" in a coauthored paper in HistoCrypt2024.)
In the nomenclature, alphabetically arranged words/morphemes/lexemes are given three- or two-letter codes arranged in the reverse alphabetical order of the initial letters (with an exception of K, which is placed between L and M). Use of the 20 different initials for the codes in Ko.1 is the richest, which is directly linked to the size of the nomenclature (p.120).
I reconstructed the same cipher from letters of Juan Perez (September 1527, see another article) and Juan Manuel (March to June 1522, see another article). While Juan Manuel's use of this cipher was not found among Kolosova's samples (p.107), he actually used this cipher (at least from 7 March 1522). Considering that the latest use of Ko.2 in Kolosova's list is 13 March 1522, I think Ko.2 was an official cipher that was replaced about this time (cf. p.107, 156).
My reconstruction may complement somewhat Kolosova's results. For example, while she did not have enough materials for identifying representation of numerals (p.122), I think they are arranged as roman numerals in their alphabetical positions for X (at least from x to xxxix).
(p.189, Substitution alphabet: p.192, Nomenclature: p.223)
This was used by Fernand Marín, Abbot of Nájera, paymaster and treasurer of the imperial army. He wrote more than a hundred letters in cipher (p.190).
There was an older cipher, but available specimens (a few sentences in a single letter) are too scanty for analysis (p.191, 432).
This cipher features four-letter codes to represent cardinal numbers. The first two letters fr is an indicator of numerals. The fourth letter (d, f, g, h, l, m, n) indicates one of the groups of five (1-5, 6-10, 11-15, ...) and the third letter (a, e, i, o, u) indicates a specific number among the five in the group. (In a way, it is like a base-5 system.) So, fred is 2, frah is 16, and so on (p.210-211).
(p.227, Substitution alphabet: p.227, Nomenclature: p.249, Specimen: p.230)
Only four letters from Pope Adrian VI to Charles V between September 1522 and March 1523 are available. They are in Spanish and the vocabulary in the nomenclature of this cipher is also in Spanish. When Adrian was a tutor to the future emperor in the Low Countries, their language was French, but Charles learned acceptable Spanish as required by the Castilian cortes when inheriting the Spanish kingdoms (Wikipedia). It is understandable that they used Spanish in official correspondence.
The cipher between Charles and Adrian has many codes for formative morphemes (p.237-238).
Of the four letters analyzed, at least no.29 is printed in Correspondance de Charles-Quint et d'Adrien VI (Internet Archive).
no.27 (p.634) 30 September 1522 (Two letters from the same date (one in Spanish and the other in French) are printed on p.122 and 125 in Correspondance.)
no.28 (p.636) 21 November 1522 (without decipherment) (A letter from the same date is printed on p.133 in Correspondance.
no.29 (p.638) 22 November 1522 (without decipherment) (The full decipherment is printed on p.137 in Correspondance.)
no.31 (p.642) 2 March 1523 (Not printed in Correspondance.)
It is known Adrian and Charles used cipher as early as 1520 before Charles left for Spain (see another article). It would be interesting to find out whether the cipher was provided by Adrian in Spain or was arranged beforehand. I think the full cipher like this was from Spain because of its similarities to the ciphers of the Catholic Monarchs.
(p.251, Substitution alphabet: p.252, Nomenclature: p.277)
Charles de Lannoy was the viceroy of Naples since 1522 and commander of Imperial armies in Italy since 1523.
(p.282, Substitution alphabet: p.284, Nomenclature: p.307)
The three-(or two-)letter codes in the nomenclature has only six initial letters (z, y, x, t, s, p). While this results in less extensive vocabulary than others, this can avoid alignment between initial letters between plaintext words and codes representing them. Such an alignment is a weakness when the initial letters cover all the letters. (p.297)
While I was aware of three ciphers of Lope de Soria (another article), none of them matches this cipher with the Emperor. This is the main cipher of the two used with the Emperor (the other is Ko.16) (p.284).
Kolosova points out similarities of Ko.6 (Lope de Soria's), Ko.7 (Lope Hurtado's), and Ko.8 (Marino Caracciolo's) (p.291-295,300,312-313,316-317, 337-338, 343-345).
Kolosova assigns diez (meaning "ten") to the code "per" and veinte ("twenty") to "pir". Probably roman numerals (x, xx) were used in the original in view of the alphabetical ordering.
An independent reconstruction is given by Galende Diaz (1992) (p.499), which I call Lode de Soria's Cipher 2 (another article).
(p.309, Substitution alphabet: p.312, Nomenclature: p.333)
Lope Hurtado used this main cipher as well as another less complex one (Ko.10) with the Emperor (p.309-310).
As with Ko.6, the initials of the nomenclature codes are limited to six (z, y, x, t, s, r).
The nomenclature includes some syllables, albeit not fully developed (p.321): mo mu na so su lo le no ni. Of these, mo, mu, na, so are not words in Spanish. I'm inclined to think the original cipher had full syllable representation but some symbols simply did not occur in the letters studied.
(p.335, Substitution alphabet: p.338, Nomenclature: p.348)
Marino Caracciolo (Wikipedia) served the Sforzas (I wonder whether he had a chance to learn Milanese ciphers) and then Pope Leo X, whose interests were contrary to the Habsburgs'. Upon Leo X's death, still remaining a nuncio under the new Pope Adrian VI (who was personally close to Charles), his duties became pro-imperial. In the spring of 1523, he was sent to Venice for the Emperor. From 1526, he worked in Milan with Antonio de Leyva and later was made Governor of Milan (p.336-337).
The letters studied are no.44 (1523) from Rome, no.55 (1524) from Pizighitone (in Lombardy), no.56 (1524) from Venice, all in Italian. Kolosova says the cipher includes both Italian and Spanish (p.335; che is Italian; dinero, ejército, tiempo, Su Sanctidad are Spanish). Given the specimens used in reconstruction are Italian, it is wondered how she can know the nomenclature included Spanish. (As far as pure nouns are concerned, there is a possibility that a word in the code in one language was used in an epistle in another language.)
There are similarities in three-letter codes between this and Lope de Soria's cipher (Ko.6) (there are cases in which the last two letters of three-letter codes are the same to represent the same word) (compare p.307 and 348; cf. p.416, p.317-318).
Outside the scope of this thesis, Caracciolo's letter to the Emperor (1537) from Milan uses a completely different cipher, not deciphered (DECODE 9966). It employs symbols with superscripts (see another article).
(p.350, Substitution alphabet: p.351, Nomenclature: p.380, Specimen: p.353)
This corresponds to the cipher I reconstructed independently (another article) from different letters in a source partly overlapping hers (my article mentions (but did not use) the letters from 22 March and 27 April 1522, which are no.19 and no.23 in her list).
Kolosova points out Ko.9 has a numeral system similar to Ko.3 but without the indicator. I'm inclined to think they are basically arranged at the location of "x..." in roman numerals. The numerical order was adopted rather than alphabetical order just for convenience (what is called "serial features" by Kolosova).
For numerals under ten, while she considers there is a "complex system of numerals" (p.380, 368), if we spell numerals instead of using Arabic figures, they nicely fit the natural positions in the alphabetical sequence of the nomenclature: 3=tres (fug), 4=quatro (lom), 6=seis (ged), 8=ocho (mi).
Sixteen letters are used as the initial letters of the three-(or two-)letter codes: z, y, x, v, t, s, r, p, n, m, l, h, g, f, d, c. This is somewhat similar to the Adornos' cipher (Ko.11).
(p.386, Substitution alphabet: p.388, Nomenclature: p.405)
This is a simpler second cipher of Lope Hurtado (cf. Ko.7). Many symbols in the substitution alphabet are taken from Alonso Sanchez' cipher (Ko.9), as shown in the table on p.390-391 and discussed thereafter in detail.
Kolosova discusses the "coincidence" between Ko.7 and Ko.10, but it would also be interesting to study the difference in use of Ko.7 and Ko.10. We know both were used in Lope Hurtado's letters to the Emperor. So it is wondered whether the difference was in time (I could not find in the thesis which letters used Ko.7 and which used Ko.10. At least, Kolosova confirms on p.432 that the two ciphers were used at the same time.), topic, secretary, or any other. Kolosova aptly brings up the question and gives various possibilities on p.431-432 but the answer is "unknown."
(p.407, Substitution alphabet: p.410, Nomenclature: p.427, Specimen: p.411)
This cipher was used by Genoese nobles, Girolamo [Gerolamo, Jerónimo, Jerome] and Antoniotto Adorno. In the effort to get control of Genoa, the Emperor allied with Antoniotto and after defeating the French, he put Antoniotto as doge of Genoa (Wikipedia). Gerolamo was his younger brother (Wikipedia in Italian). While the Emperor had Lope de Soria in the embassy in Genoa, the Adornos were given this cipher for correspondence with him (in Spanish). That they were given a full cipher like this testifies to the Emperor's trust in them (p.407).
Kolosova observes that two letters from Gerolamo to the Emperor (no.15 [February 1522] and no.20 [April 1522] from Milan) do not use "markers" (nulls at the beginning or end of ciphertext) but the letter jointly sent by him and Alonso Sanchez (no.30 [February 1523] from Venice, written by Alonso Sanchez's secretary) uses "markers" very similar to those in Alonso Sanchez's cipher (i.e., Spanish words used as nulls) (p.408-409). I agree with her view that the secretary inserted Spanish words as nulls from his usual practice.
I suppose no.42 [November 1523] from Antoniotto also uses the same cipher.
Fourteen letters are used as the initial letters of the three-(or two-)letter codes: z, y, x, v, t, s, r, p, n, m, l, g, f, d. This is the same as the sixteen used in Alonso Sanchez' cipher (Ko.9) minus h and c. From the selection of valid three-letter codes (and some more requirements), Kolosova considers the Adornos' cipher was a modification of Alonso Sanchez's cipher (p.416). I also consider the selection of initial letters of the three-letter codes provides a handy way to identify a cipher used, but we may want more evidence to be sure of the derivation relation.
Ko.1 - Ko.11 above are based on the same format (a substitution alphabet plus a nomenclature consisting of three- or two-letter codes for words/word elements arranged generally alphabetically). Ko.12-Ko.16 are simpler ciphers but I think it may be because of shortage of specimens for some of them. Ko.17 is the only syllabic cipher among the 17 reconstructions.
(p.431)
This was used in a letter (in Spanish) from July 1525. I also reconstructed the same cipher from the same letter (another article).
This is basically an "alphabetic cipher" (consisting of only a substitution alphabet), with a limited number of two-letter codes for short words.
Antonio de Leyva used a more complex cipher in 1527 (Ko.17).
(p.439)
Ludovico de Montalto had additional power in Naples during the absence of viceroy Lannoy.
This cipher is reconstructed from a single letter (in Italian) dated 16 (p.99, p.831) or 17 (p.440) March 1522.
This is a simple homophonic substitution cipher, with a few codes for words (six were found in the letter studied) (p.444).
(p.446)
The cipher was reconstructed from two letters of Raffaello di Medici (Rafael de Médicis (p.822)) in Italian (one from Florence, the other from Rome) (1521) (p.831).
This is a homophonic substitution cipher. A few codes for lexical units are given (p.451):
yeg (arma)
vag (con)
por (la)
mef (nostro)
lu (papa)
fug (svizeri)
My transcription of the three-letter codes above may not be accurate, but this seems to suggest the nomenclature was as full as for Ko.1-Ko.11, especially because the cipher appears to be only sparingly used in the two letters available (p.447).
(p.452)
This was reconstructed from three letters in Spanish from 1522.
This is a simple substitution cipher. The specimen includes three symbols for words/phrases (e.g., "ge" meaning gente de armas), but Kolosova considers these few instances are insignificant (p.453).
Interestingly, three- or two-letter combinations such as "por", "pu", "mor", "mer", looking like codes in Ko.1-Ko.11, are used as nulls (p.458-459).
(p.460)
Lope de Soria usually used a full cipher (Ko.6), but used this simple cipher in one letter (I regret I could not find which one). Kolosova considers this simple cipher was shared with someone (p.100), while also giving other possibilities (p.461).
There are no less than seven instances of codes in the relatively short specimen.
lod (de)
luc (duque)
cop (Génova)
cap (que)
cip (Venecia)
lih (Milán)
dus (Vuestra Magestad)
As it turned out, this is what I call Lope de Soria's Cipher 1 (another article) reconstructed by Galende Diaz (1992), which reveals a fuller nomenclature (p.499).
(p.467, Syllables: p.469, Substitution alphabet: p.477)
Antonio de Leyva used a simple cipher in 1525 (Ko.12) but used a different cipher in a letter from September 1527. This cipher is interesting in being the only syllabic cipher among the 17 reconstructions.
I also reconstructed the same cipher (apparently from the same letter). The characteristic feature of this cipher is its full representation of syllables by a base symbol plus some diacritics to distinguish vowels. Antonio de Leyva also used ciphers with such a vowel indicator system with Lope de Soria and Suarez de Figueroa. Charles V also used similar ciphers with Adrea Doria (1528), Suarez de Figueroa (1529, 1553), and Prince Philip (1545). It seems this type of syllabic cipher became common since the late 1520s (see another article).
Kolosova hypothesizes that he did not have a cipher for correspondence with the Emperor before the Battle of Pavia in 1525 (p.467). I speculate that his ciphers were not provided by the imperial chancery but it was Leyva who introduced ciphers he learned in Milan and the knowledge of syllabic cipher could have been from Marino Caracciolo above, who had served the Sforzas.
Kolosova seems to be aware that there were "many other syllabic ciphers" in the sixteenth century. She observes that Antonio de Leyva's cipher is characteristic in that syllables ending in A do not have diacritics (called "modifiers" [modificador]). She points out this could lure codebreakers into considering those symbols without modifiers were for single letters (p.471-472).
Three undeciphered letters to/from Charles V in BnF fr.3022 (f.16, f.26-28) and BnF Clair.322 (see another article) may be added to the list. If anyone finds any of Ko.1-Ko.17 solves this (which I could not), please let me know.
(15 April 2026) George Lasry provided me with his solution of BnF fr.3022 ciphers. It confirms two different ciphers (not among Ko.1-Ko.17) were used in letters from the Marquis of Gasto (Guasto, Vasto) to Charles V in September and November 1527. See another article.
| Wanruo Luo (2021), "El lenguaje cifrado de Isabel de Portugal (1530-1539)", dissertation, Universitat de València |
Luo's dissertation was planned to continue the work of Kolosova (p.4) but her methodology is diametrically different. She does not go into details of individual shapes or arrangements of individual symbols. Instead, she tries to be exhaustive (p.12) by consulting all the 48 files (legajos) from the relevant period in the Archivo General de Simancas (AGS) and examined 126 letters thus found (p.5-6). She also identified three relevant original ciphers in AGS, Est. Leg.1.1.1, a motley collection of ciphers from various times without proper labelling (p.4,7). Rather than discussing cipher structures, she presents her transcription of the three original ciphers and her reconstruction of no less than 38 ciphers with useful indexing (p.587 ff.) as well as edited letters (p.71 ff.) with detailed catalogue information (p.17 ff). She says she had to focus on Empress Isabella in the years 1530-1539 because of the huge amount of materials from the period. Still, a major part of her 38 reconstructions are ciphers not used in Isabella's correspondence.
The 38 ciphers she reconstructed (Cifrario 1 to Cifrario 38) are designated Lu.1 - Lu.38 herein.
Used from 1529 to 1538 by Isabella; Charles V; Minister of Treasury; Juan Vázquez; Comendador of Leon [Francisco de los Cobos]; Rodrigo Niño [ambassador in Venice].
A substitution alphabet with two homophones for each letter (three for vowels).
A section for double letters: LL, NN, RR, SS.
Syllables are represented by a base symbol plus superscripts 2-6.
Symbols with a dot above and superscript letters are nulls.
The nomenclature codes are two-letter codes or, in a few cases, three-letter codes.
Used from 1535 to 1538 between Isabella and Charles V (p.14) (but I do not find specimens from 1537-1538 in the thesis; p.596 gives use on 20-6-1538, but the letter is not in the index at f.587 ff.).
A substitution alphabet with two or three homophones.
A section for double letters: RR, SS, LL, NN, UU, AA(!).
Syllables are represented by either (i) a base symbol plus additional stroke(s) to indicate the vowel or (ii) Arabic numerals 32-91.
The nomenclature codes are figures 92-211 and graphic symbols, some of which look like a syllable symbol consisting of a base symbol plus additional stroke(s).
Symbols with a dot above and numbers from 212 are nulls.
Used from 1530 to 1532.
This is a general cipher naming its users: Empress Isabella, Margaret of Austria (presumably still alive when this cipher was prepared), and ambassadors in Rome (Miquel Mai, who also used a particular cipher Lu.24), England (Eustace Chapuys), Venice (Rodrigo Niño), Genoa (Gomez Suarez de Figueroa), and Milan (Prothonotary Caracciolo), and Don Lorenzo Manuel, Captain of Coma.
Luo gives Charles V; Pero Çapata; Comendador; Gutierre López de Padilla (ambassador in Flanders) as the users found in her materials.
A substitution alphabet with two homophones for each letter (three for vowels).
A section for double letters: LL, RR, NN, SS, UU.
Syllables are represented by a base symbol plus superscripts 1-5.
Symbols with a dot above or 6-10 above are nulls. This is good because superscripts 6-10 may look like vowel indicators 1-5.
The nomenclature codes are two-letter codes or, in a few cases, three-letter codes.
Empress Isabella was not just a spouse of Charles V but served as regent of Spain during the Emperor's absences.
It seems Isabella personally used cipher because, according to Luo, at least some autograph letters signed by Isabella were in cipher (e.g., p.23).
The edited letters are printed on p.69-586. According to the index on p.587-594 indicating which cipher was used in which letter, AGE Est. Leg. 1.1.1-112 was used from December 1529 to March 1533 in many letters between Isabella and Charles V. (Additions to the list may be made from the sources quoted in CSP: see another article.) This is a general cipher shared with others.
Isabella and Charles V appear to have also used particular ciphers between them. Lu. 37 was used in 1532, and AGS Est. Leg. 1.1.1-169 was used in 1535-1536.
A letter from Gutierre López (ambassador in Flanders) to Isabella from 5 November 1531 used Lu.14 (According to p.6, she may have used cipher with other ambassadors, too.).
The remaining 36 reconstructed ciphers are those used by other people.
One characteristic feature in the Spanish ciphers in the reign of Charles V is a vowel indicator system, of which Antonio de Leyva's cipher (1527) (also Ko.17) was one of the first among Spanish ciphers. Especially those in which the mapping of vowels to indicators (additional strokes, superscript figures, etc.) is not regular (e.g., Cipher between Lope de Soria and Antonio de Leyva (1528-1535?), Mary of Austria's Cipher (1546) below) are better than those with regular assignment typical in Philip II's time, but Luo's work shows the majority of ciphers in Charles V's time already employed regular assignment.
I thought "Figueroa-Leyva Cipher" (Lu.15) was irregular but her reconstruction suggests it is more regular than I thought. (I'd like to study more letters but use of this cipher is not recorded in the index (p.587).) Bertomeu Masià (2025) also reconstructed this cipher (BM2 below). Although her specimen did not use syllable symbols, she observes about Luo's table that it is not certain whether use of different base symbols in Dx, Mx, and Zx are because there was more than one series in some cases or because the typology was deliberately changed randomly to make deciphering more difficult. Some symbols in my reconstruction (e.g., for FA, GA) indicate irregularity is not limited to Dx, Mx, and Zx.
It is interesting to see that some users may have neglected using the syllable symbols, as seen in the small specimen given by Luo on p.628 (AGS. Est. Leg.1015 doc.42, but the sender/receiver and the date of this letter cannot be identified because it does not seem to be described in the thesis.) The letter of Valenzuela studied by Bertomeu Masià also does not use syllable symbols (two symbols used for DE and QUE are considered to be words in the nomenclature because they do not fit the syllable pattern).
Some may look irregular because of multiple patterns:
In AGS, Est. Leg. 1.1.1-169, about half of the syllables (Dx, Gx, Jx, Px, Rx, Tx, Zx, GLx, PLx) are formed by a base symbol plus regular vowel indicators, about half (Cx, Fx, Hx, Lx, Nx, QUx, Sx, Vx, Yx, FRx, GRx, PRx) are assigned sequential Arabic numerals 32-91, and some follow other patterns.
In Lu.7, superscript figures 1,2,3,4,5 are used for some while 6,7,8,9,10 are used for others to represent A,E,I,O,U. The vowels for BA, BE, BI, BO, BU fit neither pattern and are represented by additional strokes instead of figures.
Lu.G in Luo (2025) below is similar but may look more irregular.
A sense of irregularity because of multiple patterns is also seen in Figueroa-Philip Cipher (1543) and Imperial Cipher of 1555 from later decades. The latter may be actually equipped with multiple complete series but each series is regular. That is, the vowel representation is homophonic, which is the case for some consonants in Lu.15, Lu.23, and Lu.26. For example, in Lu.23, no less than four series of "base symbol + vowel indicators" are assigned to CA CE CI CO CU.
In Lu.27, the correspondence between vowels and superscript letters may look irregular because syllables are represented by a base letter plus alphabetically arranged letters in superscript:
| ba | be | bi | bo | bu | ca | ce | ... |
| ∂a | ∂b | ∂c | ∂d | ∂e | ∂f | ∂g | ... |
Lu.20 represents syllables with three-digit figures with a dot. The dot is not for indicating the vowel. (So, this is not a vowel indicator system.)
The majority class with regular vowel indicators includes AGS, Est. Leg. 1.1.1-112, AGS, Est. Leg. 1.1.1-197, Lu.3, Lu.4, Lu.6, Lu.9, Lu.12, Lu.13, Lu.17, Lu.21, Lu.23, Lu.26, Lu.30, Lu.31, Lu.33, Lu.38.
In Lu.6, not only syllables but also letters, nulls, and code words are also represented with symbols with superscript figures. Figures 19-23 are used to indicate vowels in syllables.
Lu.3 is interesting because apparently superscript figures 6-10 are used for nulls while superscript 2-5 are used as indicators for E,I,O,U. In this cipher, syllables with vowel A do not have indicators. In Lu.13, syllables with U do not have indicators.
In Lu.32, vowel indicators are assigned only for E,I,O. Syllables with A are represented by the base symbol. Somehow, syllables with U are assigned special symbols not related to the base symbols. (Of course, such singularity is not bad in terms of cryptographic security.)
AGS, Est. Leg. 1.1.1-169 employs Arabic figures 92-211 for many (but not all) entries in the nomenclature. Extensive use of Arabic figures (in particular three-digit figures) in the nomenclature was not so common in Spain at this time. (The nomenclature (except for very short ones) generally employed three- or two-letter codes.)
Lu.20 (used by Pompeo Colonna and Charles V in 1531) employs Arabic figures not only in the nomenclature but also for the substitution alphabet, null, double letters, and syllables.
For some ciphers, Luo assigns for J a symbol separate from I or Y. But at least for Lu.10 and Lu.16 and possibly also for Lu.12, Lu.18, Lu.22, the glyphs do not seem to support the distinction. I think J should be merged into I for these.
In Luo's list (unlike Kolosova's), B always has a separate entry from V. This is reasonable in view of the three original ciphers.
Also from the original ciphers, I think it is likely that symbols for double letters LL etc. were not part of the alphabet but were defined in a separate section for double letters as in Luo's reconstruction.
In Lu.15, the symbols for S and X seem to be the same. (I have also seen cases where X is represented by the symbol for S.)
In Lu.27, the letters (somehow, except for F) are represented by a base symbol ("a", "b", or "c") plus the letter itself in superscript. For example, A is enciphered as aa, ba, or ca; B is enciphered as ab or cb (probably also bb), etc. This is not so bad as one instance (Cp.30) from Philip II's time (another article), in which every letter can be represented by the letter itself!
I checked ciphers in my article with Luo's list.
"Figueroa-Charles V Cipher (1529)" does not appear to have a match in Luo's list (possibly because this is outside the span 1530-1539), while "Suarez de Figueroa's Simple Cipher" (1529-1538) corresponds to Lu.2, to which my reconstruction may add a few symbols.
"Figueroa-Leyva Cipher" (used by Suarez de Figueroa, Antonio de Leyva, Lope de Soria, Count of Cifuentes in 1533, 1536) appears to be the same as Lu.15 (used by Gutierre López, Antonio de Leiva, the Viceroy of Naples, Comendador of Leon, Count of Cifuentes in 1533-1536). Since there are symbols occurring only in either my reconstruction or hers, these may supplement each other.
"Doria-Charles V Cipher (1528)" corresponds to Lu.3, so use of the latter can be expanded from (1529-1538) to (1528-1538).
"Imperial Cipher of 1544-1554" (from Meister (1906) and Bertomeu Masia (2009)) corresponds to Lu.31 (used by the Marquis of Aguilar and Charles V in 1537-1539) (I was made aware of this by Bertomeu Masia (2025), n.11). Again, these independent reconstructions will supplement each other. The longevity of this cipher is impressive. Considering that the cipher was already used by Charles V as early as 1537, I'm curious to know in what context Diego Hurtado de Mendoza sent this to the Emperor on 17 February 1547 (Bertomeu Masia (2009), p.151).
I have been suspecting that Cp.44 and Cp.45 printed in Devos (1950) actually belong to the 1530s from the vocabulary of the nomenclature as well as from some similarities to ciphers from the period. Lu.32 (used by Lope de Soria, Charles V, Comendador Mayor de Leon, news from the Levant, the secretary of the ambassador of Venice and Diego de Mendoza in 1532-1539) appears to correspond to Cp.45 ("Don Lopez y usala Don Diego"), but there are many different symbols between the two, in particular in the substitution alphabet. Further study is needed to validate my hypothesis. I have not found a counterpart to Cp.44 among Luo's compilation.
| María José Bertomeu Masià (2025), "A proximación a los lenguajes cifrados imperiales en Florencia (1537-1538)", Estudios Románicos, DOI: https://doi.org/10.6018/ER.613261 |
Florence, long under the control of the Medici, was transformed in 1530 by Charles V into a hereditary monarch under Alessandro de' Medici (Duke of Florence from 1532). After Duke Alessandro was assassinated in January 1537 by his distant relative Lorenzino de' Medici, the title was succeeded by Cosimo de' Medici from the junior branch of the family. Filippo Strozzi, powerful Florentine banker, assembled an army and marched against Florence from France to oppose Cosimo's succession but was defeated and imprisoned. In December 1538, he was found dead in the prison but many believed his suicide was a fake.
Bertomeu Masià reconstructed four ciphers (designated BM1-BM4 herein) from the correspondence about issues concerning the succession.
Used in a letter from the Marquis of Aguilar, ambassador in Rome, to Charles V, 4 May 1537 (Spanish).
Aguilar also used other ciphers: Lu.30 (used by Juan Fernández Manrique de Lara y Pimentel, the Marquis of Aguilar, and Charles V in 1537-1539) and Lu.31 (used by many and very long-lived; see above). Bertomeu Masia points out Aguilar's expression in a letter few days later may indicate this was a private cipher with the Emperor (p.32).
The substitution alphabet includes a series of homophones that represent letters by the same base symbol ("m") plus different modifiers. (Thirteen letters of the alphabet were represented this way in the epistle studied. The original may have had a complete set.)
Syllables employ vowel indicators (called exponentes, exponente en superíndice, exponentes numéricos): 10, 11, 12, 13, 14 for A, E, I, O, U. The base symbol is "m" for Bx, "n" for Cx, and "u" for Dx. Confusingly, subsequent syllables use the same "m", "n", and "u" as base symbols but with various additional strokes as modifiers.
The same base symbols with invalid superscript figures are nulls.
The nomenclature also employs the same base symbol "m" with superscript figures above 20.
Used in a letter from Francisco de Valenzuela in Florence to Francisco de los Cobos, 20 June 1537 (Spanish).
BM2 also uses a base symbol plus a modifier to represent letters of the alphabet. This is somewhat similar to Cp.28 (1577) in the next reign (see another article).
Bertomeu Masià found this is the same as Lu.15 (see above), though there are also different symbols. So, this adds Valenzuela to the list of users of the general cipher Lu.15 and extends its in-use period by one year. She also observes that Valenzuela used Lu.8 (1530, 1534) and points out the latter might have been a private cipher.
Used in a letter from Bernardo Sanzio (later Bishop of Aquila; sent to Florence after the assassination of the Duke) to Charles V, December 1537 (Italian).
This is a simple cipher consisting of a substitution alphabet, though she warns that symbols for syllables or nomenclature might have been simply neglected by the user.
Used by Juan de Luna, castellan of the castle of Florence, 22 December 1538. According to p.28, two letters were written on this day: one to Charles V in clear (so that Cosimo de' Medici might read), accounting the death of Filippo Strozzi and the other to Francisco de los Cobos about secret details not included in the open letter.
Syllables are represented by a base symbol plus an additional stroke to (regularly) indicate a vowel. The base symbols indicating consonants are the same symbols used to indicate the letters in the substitution alphabet. For example, the base symbol for L is combined with different additional strokes to form symbols for LA, LE, LI, LO, LU.
| Sara Cascella Alcaraz (2025), "Una cifra imperial en Roma (1543)", Estudios Románicos, vol.34 (2025), DOI: https://doi.org/10.6018/ER.613381 |
Cascella Alcaraz reconstructed a cipher used in six letters between the Marquis of Aguilar, ambassador in Rome from 1536 to 1543 (Wikipedia says his embassy was from 1537 to 1540 but does seem to cite a sufficient reference), and Charles V in 1543.
She also found a draft of the cipher in the archives (AGS, Est. Leg.1.1.1-151).
The substitution alphabet has three levels.
(i) Each letter of the alphabet is represented by a symbol (a letter, an Arabic figure, or some other symbol) with a dot above or below. This may confuse a codebreaker who had an experience with a cipher using dots as vowel indicators.
(ii) The letters are assigned figures 15-37. Cascella Alcaraz observes discrepancy between the table in the archives ("31" for R) and figures used in the epistles ("30" for R) starting at R.
(iii) Each of the vowels A, E, I, O, U, Y is given one more homophone.
Syllables are represented with a base symbol plus a vowel indicator (called "significante vocálico"): 6, 7, 8, 9, 1 or 2, 3, 4, 5, 1 for A, E, I, O, U. While syllables with b, c, d, f, h, q, r, s, v fit this formula, the other syllables place the vowel indicator before the base symbol. Cascella Alcaraz points out that syllables Mx and Rx have the same base symbol but are distinguished by the position of the vowel indicator, but the base symbols seem to be different to me (MA: 6ρ, RA: p6). Graphically similar but distinct symbols are occasionally used in handwritten ciphers (Kolosova (2017), p.116).
For two-letter syllables Zx as well as three-letter syllables, the vowel indicators are dots.
The nomenclature is absent in the key in the archives, but the reconstruction reveals that names/words are represented by an alphabetical letter plus a superscript letter or digit: Alemania (bm), algunos (bd), ....
| Francisco Javier Reales Pérez (2025), "Déchiffrer l'histoire : présentation et description d'un chiffre employé par Marie de Hongrie dans sa correspondance en langue française", Estudios Románicos, vol. 34, pp.159-177 (DOI: https://doi.org/10.6018/ER.614131 ). |
Reales Pérez reconstructed a cipher used in two letters from Mary of Austria (Mary of Hungary) to (Nicolas Perrenot de) Granvelle in 1546.
The cipher has a homophonic substitution alphabet, symbols for double letters, nulls, syllables, and words.
The syllable representation of this cipher is better than many others in its irregularity.
Bx, Dx, Fx, Hx, Jx, Lx, Mx, Px, QUx, Rx, Sx, Tx, Vx are represented by "base symbol + vowel indicator" combinations, but the specific vowel indicators vary (numerals, dots, bars, additional strokes) and their mapping to the vowels is also irregular.
For Cx, Gx, Nx, it is as if the base symbol indicates the vowel. For example, the symbols for CA, CE, CI/CY, CO, CU all have a dot on either side of a base symbol, which is q for CA, p for CE, r for CI/CY, s for CO, and t for CU. Thus, vowels are not indicated by the dots (which can be said to identify the consonant C) but by the different base symbols to which the dots are attached. Similarly, Gx have two dots below in common and Nx have a bar below in common and for both, the base symbols vary to indicate the vowel. (Such distinction of vowels by different base symbols to which dots etc. are attached is also seen in Lu.G below.)
|
Eva Pich-Ponce (2023a), "Les messages secrets d'Antoine Perrenot de Granvelle à Simon Renard : la question du mariage de Marie Tudor", Thélème 38(2) (DOI: https://doi.org/10.5209/thel.90555 ) Eva Pich-Ponce (2023b), "La cifra secreta de Carlos V y la subida al trono de María Tudor", Cédille (DOI: ttps://doi.org/10.25145/j.cedille.2023.24.22 ) Eva Pich-Poonce (2024), "El lenguaje cifrado de Carlos V y el matrimonio de María Tudor", Cuadernos de Historia Moderna (DOI: https://dx.doi.org/10.5209/chmo.90692 ) Júlia Benavent (2012), "Espionaje interno en el siglo XVI. Simon Renard y Etiénne Quiclet", in F. Velasco et al. (Eds.), Cultura de Inteligencia. Un elemento para la reflexión y la colaboración Internacional (pp. 185-206). Plaza y Valdés;. (I have not been able to obtain this. The table of contents of the volume is here (pdf).) |
Pich-Ponce (2023a) discussed three ciphers used in letters between Antoine Perrenot de Granvelle, a trusted secretary of state of Emperor Charles V, and Simon Renard, ambassador to England in volumes of "Collection Granvelle" in the Besançon Municipal Library (BMB) (esp. ms. 73). (Their letters, per se familiar to historians (Papiers d'état du cardinal de Granvelle, d'après les manuscrits de la bibliothèque de Besançon (1841) (Internet Archive); CSP Spain), are in French because both were from Burgundian families.)
Three ciphers (designated PP1-PP3 below) were involved in their correspondence.
PP1 is the one used by Jean de Saint-Mauris as early as 1544, reported separately in Eva Pich-Ponce (2023b). This particular cipher attracted attention by a discovery by Pierrot, Gaudry, Zimmermann, and Desenclos in 2022 and had also been reported in my article.
It was used in more than 100 letters until 1554 by the Emperor, Saint Mauris, Simon Renard, the Bishop of Arras (Granvelle), Mary of Austria, the Bishop of Luxeuil, Francisco de los Cobos, and Ferdinand (brother of the Emperor). In particular, the Emperor used this in 1553 in his letters to ambassadors in England about the succession (Jane Grey, "Nine Days Queen", was on the throne in July 1553) and the marriage of Mary Tudor with his son Philip. Between Granvelle and Renard, this cipher was used from December 1553 to May 1554.
PP2 was only used between Granvelle and Renard, in contrast with the wide usage of PP1. It was used in Granvelle's five letters from August to November 1553. This cipher is the focus of Pich-Ponce (2023a).
PP2 is much simpler than PP1. In particular, it lacks syllables as found in PP1. Moreover, PP2 was only used sparingly. The longest passage in this cipher I spotted is "s-o-n i-n-c-l-i-n-a-c-i-o-n e-t c-o-n-t-e-n-t-e-m-e-n-t" (f.40). This is in contrast to PP1, wihch was used to encipher the substantial portion of the whole letters, apparently by a professional secretary.
Why did this PP2 have to be introduced, when PP1 and PP3 were already available? It was not that the latter ciphers were obsolete or compromised, given that they continued in use after the span of PP2. Pich-Ponce considers it was because of the Emperor's desire for strict secrecy, without even letting his closest circle know the content (p.116-117). While her explanation of the historical situation (the Emperor had to be sure of Mary Tudor's inclination before formally broaching the subject of her marriage to Philip) is very interesting, given the weak nature of PP2 (both in structure and use), there might have been other factors. (I'd like to know in what situation the Emperor (or Mary of Austria on his behalf) directly wrote to Renard (by a secretary) and in what situation Granvelle wrote to him (personally?).)
The vocabulary of the nomenclature seems to indicate PP2 was prepared for the issue of Mary Tudor's marriage. The code numbers 40 (étranger) and 45 (mariage) are used to express "mariage étranger" (foreign marriage), as opposed to marrying an Englishman (such as Edward Courtenay, a great-grandson of Edward IV). Estranger is not a word often encountered in the nomenclature.
PP3 was used from 1549 to 1554 by the Emperor, Mary of Austria, Nicolas Perrenot and Antoine Perrenot, and Renard (with a less wide distribution than PP1). This is published in Pich-Ponce (2024).
PP3 is a numerical cipher. The letters of the alphabet are assigned numbers 2-65. Interestingly, ET (55, 66, and a graphic symbol) and CON (56, 67) may have been part of the alphabet. Including ET and CON in the alphabet is a practice often seen in Italian ciphers (Meister (1906), p.171, 191, 229, 230, 231, ...).
Syllables are represented by a base symbol plus a vowel indicator (a plus sign, a bar, dot(s)).
The author says she found two original keys in the archives (AGS, Est. Leg.1.1.1, f.136, f.210) which confirmed her reconstruction of PP3 (and apparently one of the previous ones).
She notes the letter she studied (Charles V in Brussels to ambassadors in England, 31 January 1554) is extant in three copies, of which two are in cipher. She points out the two are independently enciphered in view of use of different symbols. Re-enciphering from the plaintext rather than copying the ciphertext appears to have been a common practice (see another article).
In 1555, it was discovered that Renard's secretary had been for years selling Renard's ciphers and papers to the French (another article). According to Pich-Ponce (2025), p.143, this episode appears to be described in Benavent (2012).
This may have been the main reason that Philip II decided to replace the ciphers from Charles V's time.
| Eva Pich-Ponce (2025), "Las cartas cifradas en lengua francesa de Manuel Philibert de Saboya a Simon Renard (1556)" |
This studies a cipher used in eight letters (1556) to Simon Renard from Emmanuel Philibert, Duke of Savoy, who had succeeded to the Governorship of the Netherlands (1555-1559) when Mary of Austria resigned with her brother Charles V, or Philip II.
Simon Renard was ambassador in Paris in 1556, after negotiating for the Truce of Vaucelles. The truce was unstable and only after the Battle of Saint-Quentin (1557), in which Emmanuel Philibert won a decisive victory over the French, did the war end by the Treaty of Cateau-Cambrésis in 1559.
The syllable representation of this cipher is unique. BA, BE, BI, BO, BU are represented by "ca", "ce", "ci", "co", "cu", respectively, and CA, CE, CI, CO, CU are represented by "ba", "be", "bi", "bo", "bu", respectively. Thus, syllables with paired consonants are represented reciprocally. (JA is represented by "ka", which suggests KA was also defined to be represented by "ja", though there was no instance in the letters studied.)
As was common, Renard used different ciphers with different correspondents. (The general cipher Cg.1 (1556), printed in Devos (1950), includes both Emmanuel Philibert and the ambassador in France among its users.)
Secretary Prado specifically alludes to a cipher between him and Renard (different from the one between Emmanuel Philibert and Renard).
Prado had trouble in deciphering Renard's letters and complained that Renard did not comply with the King's instructions that matters related to the states (the Low Countries) and other matters should be written in different ciphers (presumably, the one for Emmanuel Philibert and the other for Prado to be used in separate epistles).
It seems the trouble was about syllables. In saying these, Prado enciphered the Governor's letter by using only the alphabet without using syllables. Prado considered the trouble was because of the incorrect copy of the key and asked Renard to provide a copy of the syllable section, adding that the rest was unnecessary. (p.143)
| Wanruo Luo (2025), "Las cifras de Juana de Austria durante la regencia (1554-1559)" |
Joanna of Austria, sister of Philip II and widowed at 18, was called back in 1554 from Portugal to serve as regent in Spain when Philip left Valladolid for England to marry Mary Tudor.
Luo studied her correspondence during her regency (1554-1556, 1556-1559) and reconstructed two ciphers: Cipher B and Cipher G (called Lu.B and Lu.G herein). (She studied the collection of ciphers in AGS, Est. Leg.1.1.1 but, as I understand, no match was found.)
Used in 1554-1559 in letters from Charles V, Philip II, Secretary Eraso in Brussels or Philip II in Hampton Court (or "El Rey" in Rome in 1558 (Who is it?)) to Joanna of Austria, the Council of Finance, Juan Vázquez, or don Juan de Villarroel.
Syllables are represented by a base symbol plus (regular) vowel indicators. This covers not only syllables of the form CV but also CCV (e.g., fra, gli, ...). Moreover, the base symbol "dex" and "cox" seem to be used for consonants "ñ" and "rr".
Used in 1556 in letters from Philip II (in Brussels) and the Duke of Alba (in Ostia) to Joanna of Austria.
Syllables VA, VE, VI appear to be usual "base symbol + vowel indicator" combinations, which may also apply to Zx, CRx, FRx, PRx (which need more materials for confirmation). On the other hand, Bx, Cx, Dx, Fx, Gx are distinguished by the placement of dot (no dot, above, below, left, right). That is, the dot indicates consonants instead of vowels, which differ in the base symbol to which the dot is attached (5 for A, 4 for E, 3 for I, 2 for O, and 1 for U). There are some that do not fit this pattern, but it is not clear whether they were intentional variations. Similar patterns are observed for Nx, [Jx], Hx, Mx, Lx and Px, QUx, Rx, Tx, Sx.
The nomenclature includes common words represtend by code groups consisting of short letter sequences (Di, cap, effe, et, et, cen, furi, ...) or two-digit figures (16, 28, 64, ...). Within the small set of specimens, Arabic figures are only used for words beginning with Q, R, or T. Only "Vuestra Magestad" is assigned a special symbol.
|
Nicola Florio (2025), "Las últimas cifras de Margarita de Parma en su primer govierno de los Países Bajos (1566)", Estudios Románicos, vol.34, DOI: https://doi.org/10.6018/ER.613411 Júlia Benavent (2025), "Los códigos secretos de Margarita de Parma en el primer Gobierno de Flandes", Estudios Románicos, vol.34, DOI: https://doi.org/10.6018/ER.613521 |
Florio reconstructs two ciphers (E, F) used in 1566 between Margaret of Parma, Governor of the Netherlands, and Philip II. Margaret was fluent in Flemish, Italian, French, Spanish, and Latin and used both Spanish and Italian in her correspondence with Philip II. Cipher E was used in her correspondence in Italian (p.78, 77, 81)
This is a general cipher used by Philip II, Margaret of Parma, and others.
This is the same as Cg.3 printed in Devos (1950).
This is a particular cipher for Margaret of Parma and Philip II.
This has a structure similar to Cp.29 (a particular cipher between Diego Guzman de Silva, ambassador in Venice in 1569-1578, and Philip II) printed in Devos (1950).
While specific syllable representations are different, the vowel indicators for A, E, I, O are the same.
Double consonants are dissimilar. Whereas Cp.29 employs D, θ, v, M with diacritics, Cipher F employs numbers 2-16. Interestingly, double letters for Cp.29 is a subset of those of Cg.3.
The nomenclature also looks dissimilar because Cp.29 mainly employ Arabic figures whereas Cipher F employs two-letter or three-letter groups. Still, Cp.29 also employs two-letter groups from Ungria ("ba") to Zante ("fa").
Similar particular ciphers may have been created from a common template.
General ciphers Cg.4=Cg.6 and Cg.5 are also similar in that sequential numerals are assigned to consonants; vowels have homophones; and the nomenclature has two- or three-letter groups.
Benavent (2025), co-director of the project "Los lenguajes cifrados secretos de las mujeres de la Casa de Austria", gives an overview of four general ciphers and two particular ciphers used by Margaret of Parma (as well as various aspects of the practice in letter writing).
The ciphers were used for three languages: French, Spanish, and Italian. (p.20; This appears to have been confirmed in her correspondence with Álvaro de la Quadra and Philip II.)
General ciphers were shared by the King, ambassadors, and other officials.
A general cipher Cg.2 (1562) (Devos' numbering; see another article) was issued on 18 December 1562. This is archived at AGS, Est, Leg, 1.1.1-214 and used in BMB Ms Granvelle 10, f.22 (Philip II to cardinal Monçon (1564), deciphered on f.28) among others (p.20).
In November 1563, Margaret of Parma acknowledged receipt of a new cipher in a letter to Philip II. It was said that the ciphers had been in use for too long (p.20).
Philip II sent out a general cipher on 12 July 1564 (p.23). On 19 August 1564, Margaret mentioned to Granvelle a new cipher from Madrid, replacing the one composed in November 1563 and intercepted by the French in late May or early June 1564 (the cipher was stolen from the embassy in France) (p.20). This new cipher is identified as Cipher E=AGS, Est, Leg, 1.1.1-215 (as I understand from p.21). The circumstances show this corresponds to what I knew as Cg.3 (1564)=AGS,PTR,Leg,52,191 (the latter also includes other ciphers). In addition to Margaret's correspondence with the King in AGS studied in Florio (2025), BMB preserves letters in which this cipher was used by Antoine Perrenot de Granvelle, Cardinal Pacheco, the Duke of Alba, Thomás Perrenot de Granvelle, Chantonnay. AGS has letters in which this was used by Tomás de Armenteros, and Gonzalo Pérez (p.23).
It appears a fourth general cipher was used from mid-1565 at least until 1567 (p.21, 23).
Philip II sent Margaret a private cipher, Cipher F, on 12 August 1566 because he felt the general cipher was not secure (p.23, 22, 20).
Back in 1564, Margaret sent him a private cipher as soon as she heard of the compromise of the cipher then in use (CSP Simancas). If its use could be verified, this will be a third cipher used by Margaret.
Margaret shared another particular cipher with Tommaso Machiavelli (her secretary) and Tomás de Armenteros (p.23, 20).