Yet another interesting comment from Rene Zandbergen yesterday (to my flying potions post) sparked off a furious flurry of bloggery here at Cipher Mystery Mansions. While browsing through a large set of online manuscripts digitized (and hosted) by the University of Heidelberg, he found Cod(ex) Pal(atinus) Germ(anicus) 597 – an alchemical manuscript where a large amount of it is written in cipher (which you can download as a 15MB PDF file). Rene writes:-

Now this is a clear example of a MS where cipher has been used to hide secrets. It leaves me with the question:

Why does the Voynich MS not look like this?

My tentative answer: the Voynich MS isn’t actually just a cipher MS. FWIW.

(–Actually, I have my own answer to this, but we’ll get to that in a minute.–)

It seems to me that (unless Augusto Buonafalce happens to know better) the literature on Cod. Pal. Germ. 597 is pretty thin: even the Karl Bartsch catalogue entry for it (marked 287 here) isn’t much use. The Ms also merits the briefest of mentions on p.355 of the 1994 book “Geschichte der deutschen Literatur von den Anfängen bis zur Gegenwart” (it’s in Google Books). None of which, however, addresses its cipher aspect… but I guess that’s my job. 🙂 So, let’s have a look at it…

The Ms commences on folio 2r with some crossed-out ciphertext above some commentary in a different hand. (Some later pages hold only a few lines of ciphertext, so it seems likely that this originally contained just the ciphertext.) And then on folio 4v, the ciphertext (interleaved with Latin and German plaintext) starts in earnest:-

cpg597_f4v

This is a basic-looking system comprising about 23 symbols, that shows every sign of being a simple (i.e. monoalphabetic) cipher consistent with its date (1426).  The cipherbet was designed not for convenience of writing (for there are numerous fiddly characters, including a blocked-in black square), but around an apparently improvised ‘personal shape alphabet’. This points not to a cipher professional (working, say, in a Chancellery) but rather to an amateur cryptographer designing his/her own ‘homebrewed’ system:-

cpg597_main_cipherbet

The letter shapes fall into three rough groups (as per the lines above):-

  • Abstract shapes / known shapes
  • Dots and containers
  • Semi-representative (aide-memoire?) shapes (hammers, spade, rake?)

But then, just as you’re getting the hang of that, a completely different monoalphabetic cipher appears (from folio 6v onwards). This looks to be a refinement of this first system… but this post is getting a bit too long, so I’ll defer discussing that to another day.

Is this a “cipher mystery”? Yes, but only a very temporary sense, for I find it terrifically hard to believe that this wasn’t picked up by one or more of the numerous 19th century German codebreaking historians and cracked in a trice (or perhaps even a millitrice). Tony Gaffney would surely munch such a light confection before breakfast. 🙂

Finally, to respond to Rene Z’s question: why does the Voynich Manuscript not look like this? I’d prefer to start by looking at what this does resemble: Giovanni Fontana’s lightly-enciphered books of secrets, which were also from very same period. This mixing of text and ciphertext also occurs in Buonaccorso Ghiberti’s copy of his famous grandfather’s Zibaldone, which has some sections in a simple cipher, most notably what Prager & Scaglia call the “secret hoist” (on folios 95r and 98r of BR 228, for which see “Brunelleschi: Studies of his Technology and Inventions”, pp.67-70). From the simplicity of that cipher (“use the previous letter in the alphabet”), I’d suggest that Buonaccorso probably copied this from an older document, one probably made in the 1430s or 1440s (Lorenzo Ghiberti died in 1455).

Remember that this was the century when paper began to become affordable, and when ordinary people began to develop their own ciphers: and although it has become fashionable to criticize the development of individualism in the early Renaissance, I think it is fair to say that the desire to keep secrets for personal / familial gain runs in close parallel with this. Ghiberti, Fontana and the author of Cod. Pal. Germ. 597 all seem vastly similar in this respect.

Returning to Rene’s initial question, then, I suspect the correct question to be asking should be: why does the Voynich Manuscript not look like any of those ciphered manuscripts?

My own answer is that it is probably because the VMs will turn out to be from circa 1460 (i.e. 20-30 years after all of the above), and its author seems to have benefitted from contact with the sophisticated code-makers in the Milanese Chancellery, who developed and refined ideas in their own cryptographic bubble. Really, the VMs is from a very specific time and place – far too clever to be early 15th century, but still strongly mindful of what earlier ciphers looked like.

A new day dawns, bringing with it a nice email from Augusto Buonafalce in response to my post on Leonardo da Vinci’s ‘x’-like abbreviation for ‘ver (as recently mentioned by Edith Sherwood).

buonafalce_verbuonafalce_ver_normal

Augusto points out that if you remove the plain diagonal line in the reflected version, what remains appears to be similar to a ‘b’… but isn’t. In the 15th century “mercantesca” script, this particular ‘b’-like shape was used to denote ‘v’: Leone Battista Alberti suggested (in his Tuscan grammar) that this shape should be more widely employed to help tell ‘v’ apart from ‘u’. Specifically, Alberti’s ‘b’-like shape looked like this:-

buonafalce_b

Now, mercantesca hasn’t really been discussed in the context of the Voynich Manuscript before (Google returns no useful hits, while even the old VMs mailing list archives appear to be silent), which is something of a shame – for if arch-Florentines such as Leonardo, Alberti, and even Michaelangelo used it, mercantesca must surely have been as close to the beating heart of the Quattrocento project as the much-touted (but very different) ‘humanist hand’.

(The ‘humanist hand’, you may recall, is an upright, formal script that was a conscious revival of an earlier script – which is why dating the Voynich Manuscript based on supposed similarities with the the humanist hand alone is so contentious.)

While the formal humanist hand was used mainly for writing in Latin, the informal mercantesca (which flourished from 1350 to 1550, peaking around 1450-1500) was used mainly for writing in the vulgar tongue: when written well, it is sometimes called ‘bella mercantesca’.

There’s a reasonable literature on this which a Voynich researcher with palaeographic leanings ought to have at least a reasonable look through.:-

  • Orlanelli, G. ‘Osservazioni sulla scrittura mercantesca nel secoli XIV e XV’, in Studi in onore di Riccardo Filangieri (Naples 1959) I, pp.445-460
  • Irene Ceccherini (Firenze): La Genesi della Scrittura Mercantesca. (summary of 2005 poster session)
  • Albert Derolez, The palaeography of Gothic manuscript books (2003)

Having said that, it is perhaps the 45 volumes of the CMD (the Catalogue of Dated Manuscripts) produced over the last 50 years that need checking here, particularly the CMDIt (the Italian section), I suspect. A proper palaeography research challenge is something I’ve been meaning to post about for a while: but that’s definitely a job for another day…

Stuff to be thinking about! 🙂

In the 1564 printed edition of his cryptography manual, Giovan Battista Bellaso included seven challenge ciphers for his readers to break, along with a set of clues: these all remained unbroken and in obscurity until Augusto Buonafalce wrote about them in 1997, 1999, and 2006 in the journal Cryptologia.

But that’s all changed now!

Tony Gaffney – who Cipher Mysteries regulars should remember from his book “The Agony Column Codes & Ciphers” (under the nom-de-plume ‘Jean Palmer’), his reading of the Dorabella cipher, and his corrections to the Bellaso cipher transcriptions – has managed to crack Bellaso Challenge Cipher #6, despite the handicap of not actually being able to read Italian. 🙂

Here’s the ciphertext in question (with Tony’s starting point highlighted), followed by a description (based closely on the document he posted to the Ancient Cryptography forum) of how he used that to begin solving the entire cryptogram. (Incidentally, if this all comes across a bit like a kind of linguistic Sudoku, it’s because that’s essentially how most non-machine code-breaking is done)…

DP QBGTA ITP LBIEE DFIIHO LI AQILIFF SO NILEECHL OMGTTIE=
CZXRC CGEDFLLIILBGGP PLBBIUNO UL QURNXSRRNB OR ACFEDFLL=
ILBFI PLACFODACU AP UHEEOI PLSGGAOLRIBLNGIBLNPE SO ROCDBCG
BU PCLICB MR RBERPUGSTSLB PLACFOEXBUBLB BPSPDXG QU BDUU
DCCAGE FCFXSFP HP MBHI LH EOMGU FSDDHEIJMG FPDHQMPDD.

Having a repeated block of four letters five letters apart implied that the cipher system involves cycling through five different cipher alphabets: and so Tony trawled through Bellaso’s clues looking “for any word that had a period 5 repetition in it ie. lontano; riteovata; lequale; etc.” When he hit the very promising-looking word consequentemente, he lined that up with the ciphertext letters with the cycle numbers beneath:-

??consequentemente??
PLSGGAOLRIBLNGIBLNPE
12345123451234512345

There’s a problem here, in that in alphabet #4 ‘G’ appears to encipher both ‘o’ and ‘m’: yet because most printed ciphers suffer from typesetter errors, Tony ignored this and marched bravely onwards. 🙂

His next two steps forward were to notice (a) that the second letter in the group shown must be ‘t’ (it occurs in cycle #2 in the same word) and (b) the final letter must be ‘i’ (because ‘e’ is its reciprocal in cycle #5 – Bellaso was fond of reciprocal ciphers, i.e. ones that perform both the ciphering and the deciphering) – so, guessing that the first letter is ‘e’, the above section of ciphertext resolves to ‘et consequentement ?i

Observing that plaintext ‘e’ appears to get enciphered as P in #1; O in #2; N in #3; and I in #5, Tony’s next angle was to rely on the five cycling alphabets’ probably having some kind of symmetry – in particular, because P O N are all a single alphabetical step away from each other, he thought it likely that the bottom half of the alphabet was shifting along by one place in each cycle. This guess let him start to fill out the 5 cycles in more detail:-

????b?e?g??? 1
????nop?????
????b?e?g??? 2
?????nop????
????b?e?g??? 3
??????nop???
????b?e?g??? 4
???????nop??
????b?e?g??? 5
????????nop?

Where next? Well, Tony now turned his gaze on a second repeated feature in the cryptogram, which appeared to be two words formed from the same linguistic root, but with a different prefix and suffix each. Did he now have enough letters to solve this? He decided to give it a go regardless:-

ACFEDFLLILBFI &
 CGEDFLLIILBGGP
??o???????????   ???o???t?????
CGEDFLLIILBGGP & ACFEDFLLILBFI
51234512345123   5123451234512
??o???t???n???   s?????tq??n??
CGEDFLLIILBGGP & ACFEDFLLILBFI
12345123451234   1234512345123
?????tq?????o?   ?s????q??????
CGEDFLLIILBGGP & ACFEDFLLILBFI
23451234512345   2345123451234
so???q???t?one   ???p?????t???
CGEDFLLIILBGGP & ACFEDFLLILBFI
34512345123451   3451234512345
?np??????q????   ???o?????q???
CGEDFLLIILBGGP & ACFEDFLLILBFI
45123451234512   4512345123451

Looking at the fourth set, he wondered if ‘t?one‘ might well be ‘tione‘, and so tried them both “as if they were the same word”. Removing the extra I from the first word yields:-

?np?????tione   ???p?????ti??
CGEDFLLILBGGP & ACFEDFLLILBFI
4512345123451   3451234512345

His original table for #3 maps ‘e?g‘ to ‘nop‘ so it seemed entirely possible that ‘F’ might encipher ‘o’: and so guessed that this word was something along the lines of the word ‘proportion‘:-

?n proportione   ??? proporti??
CG EDFLLILBGGP & ACF EDFLLILBFI
45 12345123451   345 1234512345

Working with the code-breakers’ two secret weapons (controlled mistakenness, allied with bloodyminded persistence), Tony moved forwards, safe in the knowledge that if his guesses were significantly wrong his errors would soon present themselves. How much of the five alphabets did he now have?

??r?b?efgl?? #1
??i?nopqt???
??r?b?efgl?? #2
??di?nopqt??
??r?b?efgl?? #3
????i?nopqt?
??r?b?efgl?? #4
??u??i?nopqt
??r?b?efgl?? #5
??t???i?nopq

He now moved on to the next weakest link in the ciphertext, a long group of letters (‘RBERPUGSTSLB‘) that he thought might well now be solvable with the letters he had:-

i?nu??qc????  di??e?p??ati  ??iif?o?g?q?  u?pdgrnalcp?  ?no?l?t???on
RBERPUGSTSLB  RBERPUGSTSLB  RBERPUGSTSLB  RBERPUGSTSLB  RBERPUGSTSLB
123451234512  234512345123  345123451234  451234512345  512345123451

Pleasingly, ‘distemperati‘ seemed to fit the second version (‘di??e?p??ati’): and so he proceeded with all the remaining words in the challenge cipher.

Tony’s final plaintext (parallel with the ciphertext, and the matching cycle numbers) looks like:-

della giors cre ticip rocede qualche ilcorpo nostro ecoposto
DP    QBGTA ITP LBIEE DFIIHO LI      AQILIFF SO     NILEECHL
      23451 451 23451 234512         2345123        34512345
etorganizato inpropor tione musicabe poi maicretici sono
OMGTTIECZXRC CGEDFLLIILBGGP PLBBIUNO UL  QURNXSRRNB OR
23451234-451 45123451*23451 51234512     4512345123
disproportine etdiscrdia nella musica etconsequenteoenteli nostro
ACFEDFLLILBFI PLACFODACU AP    UHEEOI PLSGGAOLRIBLNGIBLNPE SO
3451234512345 1234512345       234512 12345123451234512345
uloriin quella giorni sono distemperati etdiscordanti ilferro inogni
ROCDBCG BU     PCLICB MR   RBERPUGSTSLB PLACFOEXBUBLB BPSPDXG QU
4512345        345123      234512345123 1234512345123 3451234
sara condoi soprese della vite per petoa saratirato serumpera
BDUU DCCAGE FCFXSFP HP    MBHI LH  EOMGU FSDDHEIJMG FPDHQMPDD.
2345 512345 5123451       2345     12345 5123451234 512345123

Bellaso appears (as per his book) to be using two letter groups to stand in for common words:

  • DP — della
  • LI — mille/qualche
  • SO — no macati e sequir/quanto/ve habiamo/scritto/nostro
  • UL — vostra/poi
  • OR — perilche/sono
  • AP — della/nella
  • BU — ditto/quella
  • MR — il vostro/sono
  • QU — quella/inogni
  • HP — della/imperoche
  • LH — intutto/per

Finally, Tony notes – “I am greatly indebted to Augusto Buonafalce for his help in translating some of the words and supplying me with copies of his English translations of the books.

The plaintext refers to Bellaso’s clue #8, and discusses the well being of the body at different times, which could well refer to the theories of the Renaissance astrologer Andrea Argoli.

All I can really say is that I think this is a splendid achievement, and I wish Tony the very best of luck with the other challenge ciphers! Excellent, well done! 🙂

In a comment to a recent post on Alberti & Averlino, ‘infinitii’ asks what my recommendations would be for a Voynich Manuscript reading list… a deceptively hard question.

Apart from the direct literature on the subject (Mary D’Imperio’s “An Elegant Enigma”, my “The Curse of the Voynich”, and perhaps even Kennedy & Churchill’s “The Voynich Manuscript”), probably the best first step would always be to buy yourself a copy of “Le Code Voynich” – not for its prolix French introduction *sigh*, but simply so that you can look at the VMs’ pages in colour. The best guide to the manuscript still remains the evidence of your own eyes. 🙂

All of which is the easy, lazy blogger answer: but the kind of proper answer infinitii alludes to would be much, much harder. I should declare here that the VMs’ life in Bohemia (and beyond) strikes me as merely a footnote to the main story (though admittedly one that has been interminably expanded, mainly for lack of proper research focus).. Given that I’m convinced (a) 1450 is pretty close, date-wise; (b) Northern Italy is pretty close, location-wise; and (c) it’s almost certainly some kind of enciphered book of secrets, then the main subject we should be reading up on is simply Quattrocento books of secrets.

Doubtless there are three or four literature trees on this that I’m completely unaware of (please tell me!): but as a high level starting point, I’d recommend Part One (the first 90 pages, though really only the last few touch on the 15th century) of William Eamon’s “Science and the Secrets of Nature” (1994). Unfortunately for us, Eamon’s main interest is in Renaissance printed books of secrets. “In Nature’s infinite book of secrecy a little I can read” (Shakespeare, Antony and Cleopatra), indeed. 🙂

From there, you’ll probably have to drill down (as I did) to individual studies of single books. Virtually everything written by Prager and Scaglia fits this bill, such as  their “Brunelleschi: Studies of His Technology and Inventions” (1970) and “Mariano Taccola and His Book De Ingeneis” (1972). I recently blogged about Battisti and Battisti’s splendid “Le Macchine Cifrate di Giovanni Fontana” (1984), and that is also definitely one to look at (though being able to read Italian tolerably well would be a distinct help there). I’ve also read articles by Patrizia Catellani on Caterina Sforza’s “Gli Experimenti” (which has a smattering of cipher in its recipes), and read up on the possible origins of Isabella Cortese’s supposed “I Secreti” (which is about as late as I’ve gone). Beyond that, you’re pretty much on your own (sorry).

As general background for what secrets such books might contain, I can yet again (though I know that infinitii will groan) only really point to Lynn Thorndike’s sprawling (but wonderful) “History of Magic & Experimental Science” (particularly Volumes III and IV on the 14th and 15th century), and his little-read “Science and Thought in the XVth Century”. Thorndike’s epic books stand proud in the middle of a largely desolate research plain, somewhat like Kubrick’s black monoliths: if anything else comes close to them, I don’t know of it.

As far as Quattrocento cryptography goes, David Kahn’s “The Codebreakers” is (despite its size) no more than an apéritif to a book that has yet to be written. I found Paolo Preto’s “I Servizi Segreti” very helpful, though limited in scope. For Leon Battista Alberti’s cryptography, Augusto Buonafalce’s exemplary modern translation of “De Cifris” is absolutely essential.

What is missing? There are a few relevant books I’ve been meaning to source but haven’t yet got round to, most notably the century-old (but possibly never surpassed) “Bibliographical Notes on Histories of Inventions & Books of Secrets” by John Ferguson. You can buy an updated version with an index and a preface by William Eamon, for example from here.

In many ways the above is no more than a very personal selection of books, and one obviously based around my own particular research programme / priorities. Yet even though I have tried to cover the ground reasonably well over the last few years, there are doubtless large clusters of (for example Italian-language) papers, books and particularly dissertations I am completely unaware of.

It should be clear that I think the basic research challenge here is to build up a properly modern bibliography of Quattrocento books of secrets, and thereby to map out the larger literature field within which the whole idea of ‘the VMs as an enciphered book of secrets’ can be properly placed. Perhaps I should use this as a test case for open source history?

A huge thanks to the indefatigable Tony Gaffney who very kindly took the time recently to double-check my transcriptions (some of them derived from Augusto Buonafalce’s transcriptions) of Bellaso’s various challenge ciphers against the copies held in the British Library.

Of the twelve corrections he suggested, roughly half were typos on my part, while the remainder were places where I had transcribed punctuation-like marks (but which were instead simply marks added incidentally as part of the printing).

I’m reasonably sure that the (corrected) Bellaso cipher page here now holds a pretty close, multiply-eyeballed set of transcriptions: so what are you waiting for, go and crack them! 🙂

After I recently mentioned Bellaso’s set of seven challenge ciphers from 1564 on this blog, Augusto Buonafalce very kindly emailed me with scans of Bellaso’s three challenge ciphers from 1555. I’ve now transcribed these (as best I can) and have added them to the existing Bellaso cipher transcriptions page.

I do acknowledge that the font that my theme currently uses for “preformatted” text is too small (thanks Dennis!), but the ciphertexts are only really there to be cut-and-pasted into whatever hacky cryptanalysis package you choose. Incidentally, one neat little online crypto cipher package is John’s Javascript Secret-Code Systems webpage, which has a number of unsolved ciphertexts, such as the three “Richard Feynman” challenge ciphertexts (copied onto a Cipher Mysteries page).

A little while back, I asked Augusto Buonafalce about Renaissance cryptographer Giovan Battista Bellaso’s challenge ciphers, completely unaware that he seems to have published more articles on them than anyone else on the planet. (Shame on me for not subscribing to Cryptologia, I really ought to.)

In fact, Bellaso published two sets of challenge ciphers in his cryptography manuals: a set of three long ones in 1553 (which I don’t have copies of), and a set of seven short ones in 1564 (which I do). For me, the mystery is why nobody has cracked any of these in 450 years… compared to the Voynich Manuscript’s multilayered (and horrendously tangled) cryptography, they can’t be that hard, surely?

Here’s a link to the short page I’ve just put up on Bellaso’s challenge ciphers. Don’t forget that the “=” signs at the line-ends are almost certainly hyphens, and not part of the cipher. Good luck! 😉