If you combine the thoughts I posted yesterday (suggesting that the “o[r]aiiv” word in the top line of f67r1 might encipher “luna”) with the “or oro ror” sequence on line #2 of f15v (which would appear to be a verbosely enciphered Roman numeral, probably “CCCC”), the two would superficially seem to be incompatible. How can the Voynichese “or“-pair encipher both “L” and “C” simultaneously?

Discarding wilfully ambiguous cipher systems (such as Brumbaugh’s “convert everything to a digit and then back to a letter), the answer would be a stateful cipher system, by which I mean a cipher system which reuses the same output letters according to which one of a set of internal states it occupies. Voynich theorists typically predict that the gallows would be the main state-switching mechanism (though Steve Ekwall also asserts that “c” / “cc” / “ch” change the internal state as well – this is what all his “folding and flipping” claims specifically relate to).

Arguably the first known stateful cipher was proposed in Alberti’s De Cifris in 1467: this was a cipher disk pair where the rotor disk rotated relative to the stator disk according to an arrangement between encipherer and decipherer (typically every few words).

Now, to modern cryptographic eyes, the whole point of per-character stateful ciphers (such as Vigenère etc) is to destroy both the numerical statistics as well as the linguistic structure of the ciphertext, as they provide two layers of information that can be used to help break that text. However, this does not seem to have been the case with Alberti’s cipher, while it certainly does not seem to be the case with Voynichese, where there is apparently both visual and statistical evidence of word structure.

Yet Voynichese uses only an alphabet-sized set of characters in its cipherbet, so does not seem to be relying on a secondary codebook at all (even Alberti’s cipher disk used a secondary codebook), so one of the few ways in which it can obfuscate its output over so many pages of ciphertext is via some form of primary statefulness.

However, there seems to be no direct evidence that Voynichese uses only statefulness: rather, it gives the impression of retaining some kind of high-level linguistic structure from the plaintext, but perhaps with letter patterns disrupted within that.

To me, the likelihood is that Voynichese evolved out of what was initially a purely stateless verbose cipher, one where (for instance) “or”, “ol”, “ar” and “al” enciphered the repeated letters in Roman numerals: M C X I. The encipherer probably then hacked his/her own system (with tricks such as the space-insertion cipher we apparently see on f15v) to hide too-obvious repetitions. However, I suspect that an Arabic digit steganography hack was later grafted into the system (the a[i][i][i]v family), probably removing the need for the “I”: and that when the time came round to creating the VMs, some kind of additional stateful disruption might well have been added to this system, whereby the or/ol/ar/al pairs swapped around depending on the state… well, that’s as far as I’ve got, anyway.

Historically, the problem is that there is no evidence of any stateful cipher system prior to Rome in 1465 (when Alberti began researching his book), which doesn’t obviously seem to square with the radiocarbon dating. All the same, it’s not the first time that different forms of dating have yielded slightly different values for the same artefact, all grist for our historical mills… 🙂

It’s been an interesting day: Edith Sherwood’s Voynich website got Slashdotted – given that Cipher Mysteries picked up 4900 visitors from that tsunami of geeky clicks, edithsherwood.com itself must have had (say) 30000 or more.

And then (just now), ORF released a teaser press release for next week’s “DAS VOYNICH-RÄTSEL” documentary to their (German-language) website. So, the real big news of the day is that the Austrian film-makers are certain that the VMS is even older than previously thought (though they don’t say by how much, or in comparison to which theory). The page says that they also took a number of ink and paint samples for analysis, and examined a number of key sections under UV light for erasures / emendations (all of which is good, exactly the kind of thing I hoped they’d do).

And here’s a site which is even more specific as to the date range and place revealed by the documentary: that it was made between 1404 and 1438 (in the “flat” part of the radiocarbon dating curve, hence the tight range), and in Northern Italy (probably or certainly?). Prepare yourself for the massed onslaught of Voynich doubters to disagree…

So, might the VMs actually turn out to be by Cicco Simonetta in his early days in the Sforza roving Chancellery? Marcello Simonetta would be pleased, but it’s still early days (I thought I’d be the first to mention it)…

PS: is it just me, or can anyone else still hear Steve Ekwall saying “It’S oLdEr ThAn YoU tHiNk“?

UPDATE: see the follow-up post “Voynich Manuscript – the state of play” for more on the Austrian documentary

Having examined many historical ciphers over the last few years, I’d say that there are only a handful of ways in which individal ‘cipherbets’  (i.e. “cipher alphabets”) are typically constructed. The big fallacy is to think that people building ciphers are only concerned with a need for long-term message security, when actually there are plenty of other important short term needs they have to attend to, such as: ease of construction, usability, speed of deciphering, aide-memoires, etc. Broadly speaking, these needs express themselves in the following aspects of the cipher alphabet:-

  1. Symmetrical – where the letter-shapes are based around a geometric / symmetrical pattern
  2. Incremental – where the cipher alphabet is adapted from a pre-existing cipher
  3. Practical – where the letter-shapes are optimized for speed of writing
  4. Stylistic – to give an overall effect of looking exotic / strange / occult / ancient
  5. Mnemonic – where letter-shapes contain associative reminders about the plaintext letter
  6. Steganographic – where letter-shapes hide visual hints as to the plaintext shape
  7. Deceptive – where letter-shapes vary in subtle ways to hinder transcription / decipherment
  8. Distracting – where letter-shapes are constructed to resemble a different type of text

Apart from ‘pure’ symmetrical ciphers (such as the various pigpen and Masonic ciphers, or indeed Edward Elgar’s Dorabella cipher alphabet),I would say that most cipher alphabets tend to present a blend of only two or three of these, which you can sensibly read as reflecting the most pressing needs of the encipherer. As brief examples, you might note that many of the Sforza ciphers were primarily [incremental + practical] (and occasionally stylistic, such as the 1464 cipher for Tristano Sforza), while I’d predict that Cod. Pal. Germ. 597 will turn out to be [mnemonic + stylistic].

What, then, of the Voynich Manuscript’s cipher alphabet? Of course, the hope is that if we can classify its cipher alphabet, we might be able to “read” the needs of its encipherer.

The first thing to note is Steve Ekwall’s extraordinarily specific claim about the four gallows shapes: he asserts that these four shapes (and their four ‘ch’ strikethrough versions) specifically depict the eight folding states of the deciphering paper key – basically, that these are mnemonic. While that would make a lot of sense, debating that in sufficient detail is something I’ll take on another time.

Regardless, my position on the Voynich Manuscript’s alphabet is simply that it is a tour de force of cipher construction technique, insofar as I think you can see traces of symmetrical, incremental, practical, stylistic, steganographic, deceptive and distracting aspects (which, curiously enough, would make Ekwall’s mnemonic the only one missing from the list). Here they are in more detail:-

  • Symmetrical
    The four gallows shapes exhibit an explicit structural symmetry – one leg or two legs, one loop or two loops.
  • Incremental
    The four strikethrough gallows look to have been developed from an earlier (probably less secure) cipher system based purely on the four simple gallows. I also suspect that the “e / ee / eee / ch / sh” letter-shapes represent vowels, and that they were in some way incrementally adapted from a variation of the “dots for vowels” ciphers used by some medieval monks.
  • Practical
    The Voynich Manuscript’s letter-shapes have been consciously constructed for ease and speed of writing, far more so than typical cipher alphabets of the time.
  • Stylistic
    I would argue that the overall form of the alphabet has been designed with older (non-cipher) alphabets in mind – that is, that the stylistics of the letter-shapes was deliberately chosen to resemble an archaic (but lost) alphabet.  Note (mainly for Elmar Vogt): I do not therefore believe that the Voynich Manuscript was meant to resemble an enciphered medieval herbal, but rather that it was meant to ressmble an unenciphered herbal written in an archaic (but lost) language. I fail to see how this makes it unlikely to be smuggled past Venetian border guards… but that’s an argument for another day!
  • Distracting
    As I argued in The Curse and elsewhere on this blog, I am convinced beyond any doubt that the “aiir” and “aiiv” cipher letter groups in the VMs are specifically meant to resemble medieval page references (i.e. “a ii v” denotes “[quire] a, [folio] ii, v[erso]”), but that this is meant to distract contemporary eyes from looking in detail beyond that.
  • Deceptive
    I believe that the actual Arabic numbers enciphered by the “aiiv” family are to be read from the shape and position of the final flourish of the “v” – and that whereas the (earlier) Currier A pages used a system based on the position of the flourish, the (later) Currier B pages used a system based on the shape of the flourish. This would also point to incremental cipherbet change during the overall writing process!

There is one further one to discuss – steganographic. If you stare at the Voynich Manuscript’s cipher alphabet long enough, I contend that you will (eventually) grasp the logic underlying most of the letter-shapes (as per the discussion above). However, you are still left with a few odd “spares” (such as “4o”, “8” and “9”) that don’t fit into the symmetric families and groups described above. What is going on with them?

In The Curse, I argued (based on the statistics) that “4o” was probably encoding a word-initial abbreviation sign: what I now think is fascinating is the notion that the letter shape for the “4” might also be steganographically hiding a horizontal stroke, as an aide-memoire to the decipherer.

Similarly, I argued (also based on the statistics) that the “8” shape and the “9” shapes were probably encoding word-middle and word-final abbreviation signs (respectively): similarly, I think that these are steganographically hiding a curved half-loop at the top of each of them, the typical mid-Quattrocento sign denoting contraction and abbreviation. I’ve marked these hidden strokes in red below:-

qokedy-highlighted

Actually, I suspect the author might possibly have given a little bit of the game away on page f2r, via a slip of the pen: para 2 line 3 word 1 is “4oP9” with a curved contraction half-loop added over the “o”, which I think might well denote a contraction of “4o” + “oP” + “9”. But that, too, is another story. 🙂

All in all, I’d say that if the Voynich Manuscript’s cipher system turns out to have broadly the same degree of subtlety and roundedness exhibited by its cipher alphabet, then no wonder it has remained unbroken for centuries. It has not only the Everest of cipher systems, but also the Rolls Royce of cipher alphabets!

While browsing through the Whipple Museum’s interesting webpages on Regiomontanus just now, I was struck by a behind-the-scenes connection that might possibly lead to the source of some of the Voynich Manuscript’s images…

In 1465, Antonio Averlino (better known as ‘Filarete’) left Milan with a letter of recommendation from his friend Filelfo in his pocket, with the intention of travelling to Istanbul to work as an architect there. I have argued (from his defaced 1445 dedication on his doors on St Peter’s Basilica, see Curse p.120) that he travelled from there to Rome – it is hardly unlikely, particularly given that Vasari believed Averlino died in Rome 1469. Note also that that Averlino may well have accompanied Domenic Dominici (the bishop of Brescia) who took the beautifully-illustarted copy (now known as Vat. Gr. 1291) of Ptolemy’s Handy Tables from Brescia to Rome in September 1465 – this is the codex which Rene Zandbergen has strongly argued was some kind of visual source for the Voynich Manuscript’s zodiac ‘nymphs’.

Now… today’s particularly intriguing observation is that the highly influential astronomer / astrologer Regiomontanus (1436-1476) lived in Rome until 1467: between 1461 and 1465, he worked for Cardinal Bessarion at his palace (which was effectively a de facto Academy / humanities research centre), where he built astrolabes, sundials, etc for his patron.

regiomontanus-small

What is relevant here is that Bessarion was born in Trebizond and was a sch0olfriend of Filelfo – and so it seems extremely likely to me that Bessarion would have been one of the key people Averlino would have planned to meet in Rome. It’s also important to note that Rome circa 1465 was not the sprawling metropolis it now is: a meeting would doubtless have been arranged.

So, if you accept that Averlino was in Rome 1465, and that he would have wanted to meet Bessarion, I think it is almost inevitable that he would have met Regiomontanus at some point. I have previously noted that Regiomontanus’ ephemerides (both in print and in manuscript, such as MS Prag 742) contained information connecting the stars with agriculture: and it is well-known that his tables also detailed appropriate positions of the moon for blood-letting. However, what is perhaps even more interesting for us is what he omitted from his tables (for that truly would be a secret), and which he apparently failed to complete before his relatively early death.

The data that was was missing was a special commentary (somewhat like a Director’s Cut?) on using astrology for medicine, for human births, and for foretelling the future. It seems seem extremely likely to me that this would have been based on the sign (and very possibly the degree) of the moon, and based on earlier (probably Arabic) works, probably via one of Pietro d’Abano’s manuscripts.

Could it be that the Voynich Manuscript’s zodiac pages, with their 30-item one-per-nymph datasets, encode the same data that Regiomontanus promised (but never delivered)? And might it have been that Regiomontanus got that per-degree data from Antonio Averlino in Rome around late 1465 – or might Averlino have instead got it then from Regiomontanus?

Of course, the spooky thing here is that this is basically what Steve Ekwall said was encoded in the zodiac nymphs. But you knew that already, right?

PS: did anyone ever find an online copy of Vatican MS 1906?

Though many people with an interest in the Voynich Manuscript will have vaguely heard of Steve Ekwall and his claims that, back in June 2000, an Excitant Spirit showed him how to make a “Folding Key” to help disentangle the Voynich’s knotted ciphertext, very few have any real idea what he’s talking about. In fact, I might just be the only one. And so I thought it might be good to YouTube-ify a short film explaining what Steve Ekwall was saying. (Specifically, how his “Folding Key” works).

As far as what you’re supposed to do with it… he believes that a Voynichese gallows character tells the decipherer to fold / flip the device to that state, an EVA ‘e’ character says to advance the device to the next state, while an EVA ‘ch’ glyph says to ‘flip’ the device to the opposite state (i.e. state 1 / 2 / 3 / 4 <–> state 5 / 6 / 7 / 8). Make of all that what you will!

If you want to print out your very own folding key, there’s a Folding Key PDF on my Compelling Press website: and for those who have yet to experience Steve’s original webpages, here are links to my copy of his main web-page, and to his additional “Folding KEY 101” page (though apologies for all the dead links there!)

I suppose this is the review I’ve spent two years steeling myself for. No matter what book critics may say, reviewing other people’s books is an easy word-game to play (typically revolving around inserting themselves into the commentary): whereas putting your own writing under the same spotlight is something closer to therapy. What, with the benefit of hindsight, do I now make of “The Curse of the Voynich“?

"The Curse of the Voynich", by Nick Pelling
“The Curse of the Voynich”, by Nick Pelling

Firstly, the title didn’t work. To an avowedly rationalist commentator such as myself, a “curse” is merely a kind of game a community plays with itself when its members all willfully look away from the ball while wondering why nothing is moving. Fair enough: but the Voynich’s own mythology is so close to fiction that the word’s far stronger associations with literary curses (the Curse of Blackadder, for example) predominates. This means that people’s first reaction is normally to wonder whether the book is some kind of curious historical fiction: so, a bit of an own-goal there.

Secondly, the cover didn’t work: Alian Design did an excellent job of interpreting the brief I sent them, and produced something that was evocative and uncertain in all the ways I intended. But, again, people have a low tolerance for uncertainty: and typically “read” the cover as somehow implying that the book lacks focus. Cover art has a rigidly defined set of conventions, which publishers (even small ones) can only pragmatically subvert, not replace: the absence of a picture of the VMs on the cover (quite literally) sent out the wrong message to buyers. This was own-goal #2.

Thirdly, the editing didn’t work. Though my friend Tabby Magas splendidly subedited my clausally-complex original draft, the overwhelming pay-per-page commercial model for digital print meant that I was forced to squeeze the whole thing into under 240 pages to keep the final price under £10 – roughly a hundred less pages than the content dictated. More pictures to support the visual arguments would have been nice, but these too used up too much of my limited page budget. And so the writing suffered.

Fourthly, the content didn’t work. Even though modern historians now routinely make use of a hugely multi(ple-)media set of influences / evidences when forming their arguments and discussions, few would dare to take on the Voynich Manuscript as a subject because of the overwhelming variety of strands that would need tackling and integrating (let alone try to draw a conclusion based on such a multi-disciplinary approach). “The Curse” set out to build an entirely new research field: while it is true that many elements of “forensic codicology” had been carried out before, I was trying to bring them all together in perhaps the most concerted way yet attempted. Essentially, I was trying to do to the many historical methodologies what mechatronics did for mechanical engineering and electronics – bring them together in parallel and direct their focus on a tangible problem. But, almost inevitably, this was too ambitious a project – to do this properly would require an entire history department, not some baldy bloke in his second bedroom with a wallful of old books, no matter how persistent he happens to be.

Finally, nobody wanted an answer. People inside the Voynich research field seem blissfully content with the irascible status quo that lays upon everything like a stifling smog: feathers get hugely ruffled if anyone so much as suggests a century for the manuscript, let alone a country, town, or (heaven forfend) an individual, never mind if they try to back it up with anything approaching an argument. At the same time, few VMs outsiders have any great interest in such questions: to most people, it’s just a historical curiosity (if, indeed, it is anything at all).

I also received some hostility about my openness to Steve Ekwall’s claims: yet only three people had written anything particularly cogent about the VMs (Rene Zandbergen and Mary D’Imperio were the other two). To me, Steve Ekwall poses a greater mystery then the VMs itself: while I have a rational explanation for everything in the VMs, I have no such explanation for Steve Ekwall. All I can do is observe that his claims about what the VMs actually is do chime to a remarkable degree with what it took me years to grasp, despite the fact that he apparently has no useful art historical grasp of the object at all. And your own rationalization for all that is… what, exactly? Of course, I could (just like everyone else does) simply pretend Steve doesn’t exist: but what is there to be scared of?

No matter: probably the biggest single criticism of my book project is that I exceeded the amount that readers could accept all in one go – it was all too much, all too soon. Yet even if (as is always possible in historical research) the whole Averlino hypothesis is somehow proven wrong, I’m pretty sure I will turn out to be at least “the right kind of wrong” – looking in the right place for the right evidence for the right reasons should be nothing to be ashamed of. In time, people will doubtless catch up and overtake me, to the point that everything in “The Curse” will stop looking like some kind of mad hallucinatory multi-dimensional take on an enigmatic Renaissance curio, and instead become high historical orthodoxy. When you’re ready, I’ll still be here.

Anyway, here’s the first punchline of the day: a brief appendix to “The Curse” that you probably weren’t expecting.

Following my recent post on Giovanni Fontana, Augusto Buonafalce kindly pointed me towards a recent single-page note he wrote for Cryptologia, suggesting that a memory machine called a “speculum” (resembling a set of concentric disks with alphabets on) designed by Giovanni Fontana might well have somehow inspired Leon Battista Alberti’s famous code wheel. But how did that idea travel? In the Quattrocento, hardly anybody knew about Giovanni Fontana’s secret works – even his encyclopedia (composed around 1450) didn’t appear in print for a further century.

In my book, I argued that when Antonio Averlino left Milan in 1465, he went to Rome, and was there when Alberti was researching and writing his little book on ciphers. I further argued that Alberti’s book has a dialogue-like summary of his debate with a different cryptographer (who, like Averlino, favoured transposition ciphers over substitution ciphers), which I argued was probably Averlino. That is to say, I concluded that the two men were probably looking at revolving cipher machinery at the same time and place. In much the same way that I don’t believe three Dutchmen independently invented the telescope at the same time, I don’t believe that Averlino and Alberti both happened to invent revolving cipher machinery at the same time and place – I believe that they were at least aware of each other, if not actually working in some kind of edgy collaboration.

But how might the idea of a “speculum” have travelled to Rome? Fontana lived until 1454, probably in Padua or nearby Venice – yet we can directly place Averlino in Venice and Padua in 1450 and 1461. What are the odds that the secrets-hungry Averlino, broadly the same kind of freelance “travelling master” as Andrea Mantegna, learned of Fontana’s mnemonic wheel directly from Fontana himself in Padua, and then brought the idea with him to Alberti in Rome? In the absence of any better information, this is now what I believe probably happened.

The odds that the secretive (and secrets-obsessed) Averlino was the person behind the VMs have already been shortened, thanks to my recent discovery (from a brief mention by Lynn Thorndike) that Averlino showed off his elegant (but now lost) herbal written in the vulgar tongue in Bergamo – and if there is a better candidate for the plaintext of the VMs’ herbal pages, I have yet to find it.

So now, here’s the second punchline of the day, which is, frankly, as hallucinatory as anything I’ve encountered.

One thing Steve Ekwall repeats over and over is the VMs’ enciphered text’s reliance on the “mirror”. The problem is that Steve has no idea what that actually means – basically, what could a “mirror” be in this kind of context? Somewhat disturbingly, the Latin for mirror is “speculum“. Could it be that it is Fontana’s letter-rearranging “speculum” that Steve Ekwall has been referring to all these years? Myself, I wouldn’t really like to say – but it’s a coincidence that makes me shudder at the thought.

My final bombshell of the day is that all of this basically closes the loop for my whole research programme – that, within the limits of the evidence currently available, I feel I have performed as complete an intellectual pathology on the VMs as is currently possible, which sharply reduces my level of curiosity about it.

I’m therefore now taking a long-term break, both from the VMs and from the blog (though please stay subscribed, as I shall still occasionally post book reviews). I’ll leave my various research leads (on dating, on f57v, and on the zodiac section) open for another day, they’ll probably still be there when I return. 🙂

But all the same, let me know if you find anything good!