In two-part code, the encoding part is sorted alphabetically or by categories to facilitate looking up plaintext letters or words to be encrypted, while the decoding part is sorted in numerical order to easily locate figures encountered in ciphertext. However, such sorting in the decoding part, it would seem, is only possible for numerical code. When arbitrary symbols rather than figures are used in the code, there is no obvious way to sort them. The present article presents a specimen of two-part code among the ciphers used by Henry IV of France, the Duke of Nevers, and some others that provides a decoding part sorted by similarity of symbols.
Many ciphers at the time used symbols (as opposed to figures/numbers or letters) to represent letters, some words, etc. but only a few had a vocabulary large enough to necessitate a separate decoding table. (A cipher of which the major part of the entries is for words rather than individual letters is called a code in modern parlance.) The cipher for Laveriere (February 1593) (no.57 in BnF fr.3995) had about 80 symbols that represent names, but syllables, words, etc. were represented by figures. A cipher ("Cifre date da uno che l'ha bussate et sono della parte Franzesa, a Monsignor Illmo Montalto") with more than 100 symbols in the nomenclature is also known (Meister p.301). These do not have a separate decoding table.
Porta's digraphic cipher assigned different symbols for all of the 20*20 two-letter combinations but the symbols sharing the same first letter have more or less similarity (see another article). Providing for syllables as well as single letters was common among Spanish ciphers at the time, in which, however, syllables were formed by combining a consonant symbol and a vowel indicator or were simply represented by figures (numbers) (see another article). In these cases, a separate decoding table would not be necessary.
The only cipher known to me that has a decoding part sorted by similarity of symbols is the one used by Henry IV, the Duke of Nevers, and some others in 1593-1594 (no.60 in BnF fr.3995). It not only had symbols representing many words and names but also had a set of different symbols for different syllables.
Of about 300 symbols of this cipher (not counting the symbols for homophonic substitution), about 100 use numerals (Roman, Arabic) and about 30 use Latin letters. (By the way, the Roman numerals use "iiiixx" for "80" (i.e., "quatre-vingt"), not "lxxx" as is common today, in the ciphers in the collection of BnF fr.3995.) Sill, there are many different symbols, which makes it desirable to provide a separate decoding table.
Several symbols are defined to repeat the preceding letter.
The following is the separate table for deciphering, in which the entries are sorted according to similarity of the symbols.
Difference in the hands of the encoding part and the decoding part suggests that this cipher did not come with the decoding table from the beginning. The encoding table seems to be in the style of ciphers of the royal court (cf. no.12, no.17, no.36, no.55). On the other hand, use of borders is similar to the ciphers of the Duke of Nevers (cf. no.25 etc.). One may guess that this cipher was given by the court to the Duke of Nevers (and others) but the Duke of Nevers (or his secretary) found it expedient to make a symbol index for decoding.
A separate origin for the decoding part is supported by the bivalance of some symbols, e.g., the "y"-like symbol for "se" and "ve", the "4"-like symbols for "i" and "u", and the symbol for "na" and "je." It appears that these double assignments were not intended for polyphonicity but simply were errors in preparing the code (see another article for polyphonic ciphers).
Of course, there is no logical way of sorting the symbols. Often, it is hard to find a symbol even with this table. For example, at first, I could not find a symbol like "crossed o" meaning "des" in this table (it is in the first column on the second page).
(Further, there seem to be some errors in the decoding table. For example, the above-mentioned "y"-like symbol should be "s" in the examples I encountered, though this may be a problem of spelling convention. Although the "4"-like symbol for "u" and the one for "i" are the same in the decoding table, the enciphering table (as well as some specimens I saw) seems to distinguish these. That is, the really "4"-like symbol was intended for "u" and the similar symbol for "i" was actually designed as "∩" plus "+", though, indeed, there are many instances where these are indistinguishable. A symbol like "crossed-η" (top-left on the second page) is said to be "re" but it should be a symbol to repeat the preceding letter.)
The above-mentioned bivalence of some symbols troubled me when I first tried to reconstruct this cipher from the already deciphered materials.
The following is my partial reconstruction of the cipher.
My reconstructing work was abandoned when I found the original cipher shown in the above. Still, the process gave me some insight into its use. Some 80 symbols for substitution for single letters as well as some symbols for syllables are used often enough that after some practice, I could read them without frequently referring to the table. On the other hand, it is difficult to look up a particular symbol representing a word not used so often. Actually, even with the full decoding table, I found it hard to fully decode some ciphertext that remain undeciphered in the archives, as noted below. The reader is kindly asked to provide reading of these undeciphered texts.
This cipher is used in the following letters.
By way of context, Louis Gozaga, Duke of Nevers [also spelled as Duc de Nivernois, Nivernoys, Nyvernois], (1539-1595) was a fervent Catholic but, being a royalist, after some temporization, supported Henry IV rather than the Catholic League. (Wikipedia) After Henry IV's conversion to Catholicism on 25 July 1593, the Duke of Nevers was sent to Rome to obtain absolution from the Pope but with no avail (Baird around p.365). The papal absolution was given only in September 1595 (Baird around p.367).
BnF fr.3984 (Gallica)
(f.197) Endorsement to a letter of 24 July 1593 from Rome, stating "Coppie de la lettre escrite de Rome au Sieur de Maisses". Sieur de Maisse was ambassador in Venice.
BnF fr.3985 (Gallica)
(f.84) Jeronimo Gondi [Jerome Gondy] to Henry IV, "Planore, une poste de Laugne", 2 June 1593
Interlined deciphering..
(f.88 2nd para.) Duke of Nevers, to Revol (Secretary of State of Henry IV), 21 August 1593
Some portions in cipher (undeciphered), which appear to include words "de la [c]onversion et absolution particulieremant", "le faict de labsolution" and "le pretexte quil estoit excomuni[e]".
The Duke of Nevers was sent to Rome by the King in 1593.
(f.115 2nd para.) Duke of Nevers, to M. Revol, 27 August 1593
Some portions in cipher (undeciphered), which appear to include "reponses par Ambassadeur d'Espaigne".
(f.120) Jeronimo Gondi to M. Revol, 30 August 1593 (in Italian)
Interlined deciphering. (Both plaintext and ciphertext are neatly written.)
(f.126) Henry IV to Duke of Nevers, Melun, 31 August 1593
Instruction about the latter's mission to the Pope. Printed in Memoire de Nevers, p.492 ff. Cipher begins on f.130 (Memoire p.498).
(f.176) Duke of Nevers to M. Revol, 2 September 1593
Some portions in cipher (undeciphered), which appear to include words such as "ambasadeur", "roy avoyt", "deus dimanches", etc. The code lxxix (rome) is also used.
(f.204) Henry IV to Duke of Nyvernois, Fontainebleau, 8 September 1593
Interlined deciphering.
BnF fr.3986 (Gallica)
(f.58) Henry IV to Duke of Nyvernois, Fontainebleau, 25 September 1593
Interlined deciphering.
(f.146v) Duke of Nevers to M. Revol, Vese, 7 October 1593
Some portions in cipher, apparently undeciphered, which partly reads "demander l'absolution", "a la verite", "interest".
(f.151) Henry IV to Duke of Nivernois, Chartres, 7 October 1593
Interlined deciphering.
(f.157) Duke of Nevers to M. Revol, Coyre, 9 October 1593
Some portions in cipher, undeciphered, which partly reads "aduis", "endroit."
(f.174) Henry IV to Duke of Nivernois, Mante, 16 October 1593
Interlined deciphering.
(f.191) Henry IV to Duke of Nivernois, Mante, 22 October 1593
Interlined deciphering.
(f.198) Duke of Nevers to Revol, Desanzan, 23 October 1593
Some portions in cipher (undeciphered), which partially reads "je vous suplie de ne decorager le Roy afin quil me asiste avec ses actions a [vai?]ncre la timidite du pape a mechansete des espagnols et ligueurs".
BnF fr.3987 (Gallica)
(f.25) Henry IV to Duke of Nivernois, Dieppe, 6 November 1593
Interlined deciphering.
(f.54) Henry IV to Duke of Nivernois, Dieppe, 10 November 1593
Interlined deciphering.
(f.66) Duke of Nevers to M. Revol, 10 November 1593
Some portions in cipher, undeciphered, which appear to include words such as "tous sont de opinion que le pape veul ... asseurance de la conversion ... pour vous dire ...", "conposer avec Mr. Despernon pour la deliuranse de son frer".
(fol.143) Henry IV to Duke of Nivernois, Dieppe, 26 November 1593
Interlined deciphering.
BnF fr. 3988 (Gallica)
(fol.99) Louis Revol to Duke of Nivernois, Mante, 22 December 1593
Interlined deciphering. (The symbols are neatly written.)
(fol.119) Henry IV to Duke of Nivernois, Mante, 23 December 1593
Interlined deciphering.
(fol.143) Henry IV to Duke of Nivernois, Mante, 22 December 1593
Interlined deciphering.
BnF fr.3989 (Gallica)
(fol.3) Duke of Nevers to Revol, 1 January 1594
Some portions in cipher, undeciphered, which appears to partially read "Nyvernois du cote que je dois faire a notre retour soyt par ...."
(fol.59, 60) Henry IV to Duke of Nivernois, Mante, 28 January 1594
Interlined deciphering.
(fol.169) Duke of Nevers to Revol, 12 March 1594
Some portions in cipher, undeciphered, which partially reads "de la diuision", "discours susdit."
BnF fr.3990 (Gallica)
(fol.27) Duke of Nevers to Henry IV, 6 May 1594
Undeciphered. It can be read for the most part as follows.
See another article.
data.bnf.fr gives variants of the spelling of Nevers.