Painters also sometimes use encrypted writings, for example in their notes or as a dedication to a painting; here we are dealing with the painter Salvo, pseudonym of Salvatore Mangione, born in Sicily in 1947 and active in Turin until his death in 2015. An Archivio Salvo was created in Turin for the protection and promotion of his works.
In February 2022 I received an email from Salvo's daughter, Norma Mangione, who proposed me an encrypted text by the painter, visible in the figure on the right, on the back of one of his paintings from 1977 depicting the temple "G" of Selinunte. They had tried in vain to decipher it and were asking me if I was able to get anything out of it. Salvo dabbled in cryptographies, and had made at least one other, which they had solved like an anagram; in fact it was a simple diagonal rectangular transposition.
The first step in attacking the cipher is to do frequency analysis of the signs; it doesn't take long and the result is shown in the next figure. For a text of just $ 36 $ signs the frequency of the single sign is not very significant, for example the most frequent sign ꬴ does not necessarily stand for the most frequent Italian letter which is E. However, it is known that the percentage of vowels is very stable even for short texts, and in Italian it is about $ 46\%$, and $43\% $ for AEIO only; in Italian more than 90% of the words ends in vowel, which is of great help here. The signs at the end of the word are ꬱ ꬵ ꬴ ꬴ ꬵ ꬱ ꬾ , which are then four distinct signs: ꬱ ꬵ ꬴ ꬾ . Is it possible that these are the four vowels A E I O? By adding their frequencies we have: $ 2 + 5 + 7 + 3 = 17 $, the percentage is therefore $ \frac{17}{36} = 0.4722 \dots = 47.2\% $ a little high for the 4 letters AEIO, but almost normal for all vowels, a strong indication that the U is not present in this text, and that the other four vowels are just these.
Thereafter I was trying different permutations of the four vowels; first using the combination ꬱ = O | I ; ꬵ = A ; ꬴ = E ; ꬾ = I / O could read a ELEVATI or ELEVATO for the first word and a TRA at the beginning of the last line, but I didn't go much further.
Another significant parameter also for short texts, useful for distinguishing a mono-alphabetic cipher from a poly-alphabetic one, is the number of presences , aliases of different signs used to encrypt; for a text of 36 characters the expected number for a mono-alphabetic cipher in Italian is a little less than 15, and here the actual number is 14, in good agreement with the mono-alphabetic hypothesis.
Then I tried ꬱ = O, ꬵ = E ꬴ = A ꬾ = I we read at the end of the first line A..I.E which suggested to me ATRIDE plausible perhaps, since it is a painting in ancient Magna Graecia. Replacing T R D elsewhere, the first word of the third line was TRE but otherwise I didn't go much further
A widely used and often effective method is to look for a crib ↑; that is a word that for some reason should be very likely present in the message.
In this case the crib came from Norma, who wrote me an email with some information on the context, in particular that the painting was intended for Achille Delpiano, a mechanic friend of the painter and who had been commissioned of a job on his car. She also wrote to me that she had given up trying to decrypt the text, having realized that there were no doubles, and that therefore the word Achille with that double L, could not be there.
ꬴ ꬰ ꬴ ꬹ ꬾ ꬳ ꬱ 𐂏 ꬴ ꬳ ꬷ ꬾ ꬰ ꬵ ꬷ ꬵ ꬰ ꬴ ꬽ ꬲ ꬵ ꬻ ꬴ ꬼ ꬺ ꬴ ꬳ ꬷ ꬵ 𐂏 ꬸ ꬱ 𐂏 ꬴ ꬰ ꬻ ꬵ ꬰ ꬾ
Paradoxically, however, that mention of doubles was the crib that allowed me to begin to break the cipher. Indeed, a good rule in classical cryptography, clearly unknown to many amateur cryptographers, is to eliminate the doubles that would be a formidable help for the cryptanalyst. They are usually simply eliminated, for example by writing Achile instead of Achille. L.B. Alberti and other cryptographers actually advised ignoring the rules of spelling! In many Renaissance ciphers, instead, specific ciphers were used for the various doubles, one for BB, one for CC, etc.
AL AMICO ACHILE HELA ..E.A..A CHE .O AL.ELI
Assuming that Salvo had chosen to ignore the doubles, instead of ATRIDE it was now easy to read ACHILE, and substituting in the rest of the line I got: ALA.I.O ACHILE. ALARICO? Not very plausible, while it was reading AL AMICO ACHILE↓↓, better restoring the doubles: ALL'AMICO ACHILLE, an excellent start, very plausible. By substituting the ciphers already resolved in the rest, the text on the side was obtained.
ALL AMICO ACHILLE NELLA SPERANZA CHE .O ALREL[L]I
Now, the word on the right of the second line could be SPERANZA (Hope), which admitting an oversight by Salvo in the initial H, whose cipher is very similar to the one of N. On the right, the resulting decryption.
ALL AMICO ACHILLE NELLA SPERANZA CHE VO AL RELLI
ALL AMICO ACHILLE NELLA SPERANZA CHE LO APPREZZI
This possible solution, sent to the Salvo archive, does not at all convince the family of the painter-cryptographer; This time Cristina, Salvo's widow, after reading this last solution, 45 years later, tells me that Salvo had no interest in rallying, and that she begins to remember something, agrees on the first two lines but writes that she has a vague memory that with that inverted green triangle her husband meant any double and therefore she proposes the solution visible on the right, which ends with "LO APPREZZI", a double P and a double Z.↓↓.
In cryptographic jargon, such a thing, a sign that encrypts many different letters, in this case pairs of different letters, is called polyphon. Polyphons have played a marginal role in the history of cryptography, for the simple reason that the decryption is no longer unique and can give rise to ambiguity. The only case of professional use of polyphons that I know, is that of the polyphonic ciphers of Giambattista Argenti and his nephew Matteo Argenti secretaries of the ciphers of the Papal State at the end of the sixteenth century. An example is the cipher for the 1585 inquitore of Malta
The wide-ranging polyphone used by Salvo, considering that in Italian all consonants can double, actually leaves the way open to many ambiguities. Here, for example, the last word could also be ATTREZZI or AFFRETTI, which give correct solutions from the lexical-syntactic point of view, and after all also at the semantic level. Even the first part ALL'AMICO could be AFFATICO, since the T does not appear in the main solution.
Ultimately, the solution proposed by Cristina is the most plausible and probable, precisely because it comes from a person who knew the author well, who at the time had seen the solution, then forgotten and now resurfaced in memory.
Admitting that the most probable solution is the one with a polyphone for doubles, with the cryptogram ending in "LO APPREZZI" the key of the cipher is given by an alphabet necessarily incomplete because some letters do not appear in the text, and the doubles are encrypted with a unique sign.
This cryptogram presents several interesting points to keep in mind when trying to solve short cryptograms, the work of amateurs, first of all the fact that more acceptable solutions can be found.
----raw .. ..... ...... .... ........ ... .. .. .... AD AVITO ATRIDE .EDA SPERANZA TRE .O AD REDI AL AMICO ACHILE NELA SPERANZA CHE DO AL RELI ----admitting a double L ALL AMICO ACHILLE NELLA SPERANZA CHE VO AL RELLI ----admitting a unique polyphone for all doubles AFFATICO ACHILLE NELLA SPERANZA CHE VO AL RELLI AFFATICO ACHILLE NELLA SPERANZA CHE DO AL RELLI AFFATICO ACHILLE NELLA SPERANZA CHE LO APPREZZI ALL'AMICO ACHILLE NELLA SPERANZA CHE LO ATTREZZI AFFATICO ACHILLE NELLA SPERANZA CHE LO AFFRETTI ALL'AMICO ACHILLE NELLA SPERANZA CHE LO AFFRETTI
AL AMICO ACHILE to the friend Achilles HELA ..E.A..A in the hope CHE .O AL.ELI that ?? ??????
ALL AMICO ACHILLE to the friend Achilles NELLA SPERANZA in the hope CHE VO AL RELLI that I go to the Rally
ALL AMICO ACHILLE to the friend Achilles NELLA SPERANZA in the hope CHE LO APPREZZI that he appraises it