I succeeded in solving a cipher in a letter to Emperor Charles V from Suarez de Figueroa, Spanish ambassador in Genoa, dated 24 May 1529 (PARES). At first glance, the ciphertext looks very complex with superscript and subscript numerals and other diacritics. However, many Spanish ciphers in the sixteenth century used a vowel indicator system, whereby diacritic signs indicate vowels. Indeed, such a cipher was used by Admiral Andrea Doria of Genoa in the preceding months (called the Doria-Charles V Cipher (1528) in another article). If the ambassador's cipher is based on a similar system, codebreaking should not be very difficult. After all, a vowel indicator system separates vowels from consonants, which greatly facilitates the task of codebreaking. Moreover, the ciphertext leaves some word divisions.
The present article describes how I solved this cipher, which I call the Figueroa-Charles V Cipher (1529). Although cryptanalysis against a Spanish cipher with vowel indicators is already described in Adler et al., "Reading Encrypted Diplomatic Correspondence: An Undergraduate Research Project", Cryptologia 32:1 (2008), I believe an additional specimen is of some interest. Codebreaking is an art that cannot be covered by general principles alone.
One first notes that superscript/subscript numerals are limited to 2, 3, 7 (subscript) and 2 (superscript). I assumed these were vowel indicators and, to see which numeral stood for which vowel, I counted the frequencies of these numerals occurring on the first page (using a Japanese tally symbol "正"). I thought the frequent subscript numerals 2 and 3 were either E or O, the least frequent one (subscript 7) was U, and the superscript 2 was I. As to A, by analogy from the Doria-Charles V Cipher (1528), I was feeling that the two dots below a letter might indicate A. As it turned out, these were all correct. But I was not sure enough to pursue these hypotheses.
A dot above a letter is used only in sequences at the beginning and the end of ciphertext and may well indicate a null.
There are some repeated patterns such as m2fi3o3n2h2f and o3h2d2sd2na (cipher symbols herein are approximate representations). I searched for words with these syllable structures. To take the second pattern as an example, o3, h2 and d2 would be Consonant+Vowel and s, n, and a would be single letters. Moreover, d2 is repeated. I found fifteen words matching this pattern: BABOSEASEIS, CACAREAREIS, COLOREAREIS, DISOCIACION, MANOSEASEIS, NEGOCIACION, PAJAREAREIS, PAYASEASEIS, RAPOSEASEIS, RUMOREAREIS, SABOREAREIS, SENOREAREIS, TARAREAREIS, VITOREAREIS. Of these, NEGOCIACION seemed plausible. (By the way, the word also gave a breakthrough when I broke an Italian cipher earlier this year.) Again, NEGOCIACION turned out to be correct in the end, but I could not be sure at the time, partly because it was not consistent with hypotheses obtained from the first pattern.
I once gave up solving this myself.
On 17 December 2018, I had to be on train for an hour, during which I gave a look at the ciphertext again. The above identification NEGOCIACION would give identifications:
o3(NE) h2(GO) d2(CI) s(A) d2(CI) n(O) a(N).
That is, d* reads C*, *2 reads *O, and a reads N. The word following this pattern in the ciphertext is "d2a", which, with the above identifications, can be read as CON. (This alone demonstrates the weakness of the vowel indicator system.) It is further followed by +14, obviously a code symbol for some proper noun. "Negotiation with somebody" completely made sense. Moreover, the frequency of consonants seemed right.
I now felt I could pursue these identifications further.
Then, I found a pattern on the second page: rav3m2h3ad2B, which partially reads:
r(*) a(N) v3(*E) m2(*I) h3(GE) a(N) d2(CI) B(*).
Although I do not know Spanish, loanwords of Latin origin in English allowed me to recognize Spanish words. In this case, it was not difficult to guess INTELIGENCIA, which revealed further symbols.
INTELIGENCIA is followed by CON, which in turn is followed by:
B(A) p(*) h7(GU) o..(NA) f(*).
(By this time, I felt sure enough of my hypotheses for the vowel indicators "7" and "..".) Here, I could see a word ALGUNAS, one of the few Spanish words I had learnt in working with Spanish ciphers.
Now, the first pattern above could be partially read:
m2(LO) f(S) i3(*E) o3(NE) n2(*I) h2(GO) f(S),
which suggests a word ENEMIGOS. I was not sure of the consonant for i3(*E) but soon the regularity of the cipher (see below) indicated it should be HE. (Google search returns many instances of "los henemigos" in Spanish.)
Now, it was a matter of time to reveal the remaining symbols. The word INPORTANCIA confirmed a symbol for I and allowed for identifying P and R. The name ANDREA DORIA gave D. The word PORQUE gave QUE. Thus, analogy from English, historical background, and the least knowledge of Spanish words all played a part in the deciphering.
The reconstructed cipher is shown below. As expected, it is similar to the Doria-Charles V Cipher (1528) even in the regularity in assignment of base symbols for syllables (e.g., "d" for "ca, ce, ci, co, cu"). Use of superscript numerals is an additional feature but it is pretty much obvious that the sole superscript "2" would be nothing more than another subscript numeral.
The following is my preliminary reading of the letter. (Italics are clear text in the original. Brackets [ ] indicate yet undeciphered cipher symbols.)
The crossout in the manuscript ciphertext indicates it was deciphered on a separate sheet. The plaintext may be found somewhere in the archives.
S. Tomokiyo, Spanish Ciphers during the reign of Emperor Charles V
S. Tomokiyo, Spanish Ciphers during the Reign of Philip II