Category Archives: Voynich Manuscript

Statistical and cryptanalytical analyses tend to assume that ciphers will fit one of a small number of well-known and well-researched pigeonholes (e.g. Vigenère, autokey, etc). Unfortunately, this kind of “backwards attack” can often be stopped dead if the encipherer includes one or more additional steps sideways, unless the backwards attacker happens to be cunning or lucky enough to reconstruct those tiny steps.

But as they knew in Bletchley Park, it is sometimes possible to “forwards attack” cryptograms. There, a “crib” was the name BP codebreakers used to describe where you already had the plaintext, typically obtained by decrypting the same message enciphered using a different cipher system: having this would help the code-breakers reconstruct daily settings for the second cipher etc. Just so you know, this is precisely why you should never forward a received (and deciphered) message word-for-word using a different cipher, a lesson many WW2 code bureaux stubbornly failed to learn.

Similarly, the idea behind my “block paradigm” methodology is that if we can use secondary historical clues to determine the plaintext from which a given section of ciphertext was derived, we stand a reasonably good chance (I think) of reconstructing the cipher forwards from there. You can therefore think of it as a high-level “historical crib”, where the plaintext is reconstructed via in-depth research rather than by breaking a parallel cipher.

At the very least, this whole process could very possibly yield a completely different class of problem to solve, which in the case of the Voynich Manuscript shouldn’t be a bad thing, given that a century’s worth of backwards attacks has been largely unproductive. 🙁

But what might the plaintext for the Voynich zodiac look like? Would we even recognize it if we had it in front of us?

The Voynich zodiac section

In the same way that many people have long suspected that the Voynich Manuscript’s “Herbal” section(s) probably contains plant and/or remedy descriptions (albeit secret, valuable or unexpected ones), there has long been a strong – yet untested – historical hypothesis about what the Voynich zodiac section might well contain, which is simply this: per-degree astrology. This is because each sign seems to be divided into 30 elements (29 in the case of Pisces, though this may possibly have simply been a slip of the quill), and there are 30 degrees in each zodiac sign (i.e. 12 x 30 = 360).

The modern history of per-degree astrology is something I covered here before: it moved back from Marc Edmund Jones (1925) to the nineteenth century astrologers “Charubel” (who claimed he channelled his per-degree symbols) and “Sephariel” (who claimed that he copied his from “La Volasfera”, supposedly a Renaissance book by Antonio Borelli / Bonelli [did he mean Guido Bonatti?], but this has never turned up).

It’s often written that Western medieval per-degree astrology arrived from Arabic sources via Pietro d’Abano (while he was in Spain during the 13th century). Heidelberg has a 15th century German translation of his work in MS Cod. Pal. Germ. 832 (“Regensburg, nach 1491”), which Rene Zandbergen mentioned in a comment here back in 2009. (If you look at fol. 36r onwards, you can see a few lines of text for each of the thirty degrees in each of the signs in turn, along with some rather jaunty miniatures.)

Prior to the Arabs, you can doubtless trace all this back to the original Indian sources (Diane O’Donovan pointed to the encyclopaedia-sized “Brihat-Samhita” by Varahamihira), but taking things back that far falls way beyond my paygrade, so I’m not going to attempt it in this post. 🙂

However, if you take the time to read Chapter XII of volume III of Lynn Thorndike’s “History of Magic and Experimental Science”, you’ll see that another medieval writer famously wrote on per-degree judicial astrology: and this is where my search began.

Andalò di Negro

Andalò di Negro (fl. first half of the 14th century) was a noble from Genoa. Boccaccio, who he famously taught “in the movements of the stars”, noted that “since [Andalò] had traversed nearly the whole world, and had profited by experience under every clime and every horizon, he knew as an eye-witness what we learn from hearsay” (De genealogia deorum, XV, 6, quoted in Thorndike, p.195).

One of Andalò’s works (“introductorium ad iudicia astrologie”) that discussed per-degree judidicial astrology was of particular interest to me. So, back in 2009, I managed to get some working photocopies of it courtesy of the Warburg Institute’s Photographic Collection: these were of the two known documents listed further below.

Interestingly (and unlike the Pietro d’Abano-derived Heidelberg manuscript), the key feature that seems to oddly parallel what we see in the Voynich Manuscript’s zodiac section is that these two documents contain only a small amount of data per degree (though admittedly arranged in columns of a table rather than in the form of nymphs, stars and labels).

The first document is at the British Library: I managed to get a look at this in person, kindly thanks (if I recall correctly) to a letter of introduction from Dr David Juste, who was then a historian of astrology at the Warburg Institute. One unusual feature was that a few key parts of the tables were highlighted in different colours, something that wasn’t at all apparent from the black-and-white photographs taken for the Warburg in the (I guess) 1920s or 1930s. (Sadly, the colour notes I took at the time have long since disappeared).

* BL Add. MS 23770 (BL: “14th century”, Warburg: “circa 1350”)
http://searcharchives.bl.uk/IAMS_VU2:IAMS032-002097417 – “Letter of introduction required to view this manuscript”
1. “INTRODUCTORIUS ad iudicia astrologie co[m]positus ab A[n]dalo de Nigro de Janua;” with paintings of the signs of the Zodiac, the planetary Gods, etc., ff. 1-44.
Aries (8r), Taurus (9v), Gemini (11r), Cancer (12v)
Leo (13v), Virgo (15r), Libra (16v), Scorpio (17v)
Sagittarius (18v), Capricorn (19v), Aquarius (20v), Pisces (21v)
You can see monochrome thumbnails of these twelve images on the Warburg Institute’s Photographic Collection.

In the case of the second document, since 2009 Cod. Fonds 7272 has been placed online and made downloadable by BNF. As a result, I can include links directly into the Gallica pages for you (which is nice).

* BNF Cod. Fonds Latin 7272
http://gallica.bnf.fr/ark:/12148/btv1b8452771j
Aries (112r), Taurus (114r), Gemini (115v), Cancer (116v)
Leo (118v), Virgo (119v), Libra (121r), Scorpio (122v)
Sagittarius (124v), Capricorn (126r), Aquarius (127v), Pisces (129r)

Here is the Aries table as it appears in BNF Cod. Fonds Latin 7272, courtesy of Bibliothèque nationale de France via Gallica:

By way of a guide, fol. 22v of BL Add. MS 23770 explains (says Thorndike, p.192 n.5) Andalò di Negro’s “fivefold distinction of degrees within the signs: 1, masculine or feminine; 2, lucidus, tenebrosus, fumosi, or vacui; 3. putei; 4. azamena (like the putei, to be avoided); 5, augmentates fortunam.”

However, I have to mention at this point that according to Boncompagni’s (1875) “Un Catalogo dei Lavori di Andalò di Negro” (an offprint taken from “Bullettino di Bibliographia e di Storia delle Scienze, Matematiche E Fisiche”, Tomo VII – Luglio 1874, and for an original of which I paid a load of money several years ago but which is now available print-on-demand from Kessinger *sigh*), there might possibly be a third copy still floating around.

Boncompagni (pp.54-55) mentions that an 1834 alphabetical index of the Biblioteca Altieri di Roma published by Federico Blume lists: “de Nigro, Andali de Ianua, Introductorium ad iudicia astrologiae, Fogl. membr. V.E.5”. Moreover, Emilio Altieri’s index to the Biblioteca Altieri (car. 11a, recto, lin.2) reads: “Andalus de Nigro de Janua, de Astrologia, Pil. 13, Lett. A. Numo. 5”. Yet according to a 1690 index, “Il detto Altieri non possiede ora alcun esemplare manoscritto d’alcun lavoro di Andalo di Negro“, so it seems that it had already disappeared by then.

That sums up the known versions of this work tolerably well: but what might these tell us about the Voynich zodiac? Obviously, that’s a good question something I’ll have to leave for a follow-on post…

A “research tree” is the term I like to use to describe a whole group of evidence / artefacts / phenomena / ideas that are linked together in non-chain-like ways. The term is particularly relevant to unsolved cipher mysteries because you almost always start by not knowing where in particular research trees your cipher fits (if it even fits at all, which they typically don’t).

So, to both recap and expand slightly:

(1) I suspect that the Voynich Manuscript’s Quire 20 contains a sizeable collection of encrypted recipes.
– – Though an old suggestion, we now have a lot of secondary analysis to help reconstruct its original page order.

(2) I suspect we now know enough to be able to match Quire 20’s original structure with that of unencrypted recipe collections.
– – This matching trick is what I call the “block paradigm” approach to cracking historical ciphertexts.

(3) The most likely languages for the plaintext are Italian, Latin, and French.
– – The ‘michitonese’ handwriting appears to contain some Voynichese, and looks to have come from Savoy.

(4) I believe we can eliminate Latin as the plaintext language.
– – This is because Voynichese’s ‘8’ and ‘9’ characters appear to function as ‘contraction’ and ‘truncation’ shorthand tokens, making them essentially incompatible with Latin (where word endings hold a large amount of semantic content).

(5) My working hypothesis is that the plaintext is in Italian (Tuscan).
– – This is because there are a large number of Italian herbals, but very few French herbals.

(6) The various reliable dating evidences we have suggest that this was written between 1440 and 1470.
– – (…don’t get me started on this, or we’ll be here all night.)

(Feel free to disagree with any of the above! I’m not telling you what to think, but making clear the constraints I’m using to guide my own search.)

As a result, I’ve been looking for 13th / 14th / 15th century recipe lists written in Italian. My current hunch is that Quire 20’s plaintext might well be something close to BNF MS Latin 6741 – Jean le Bègue’s collection of paint, colouring, and gilding recipes.

Hence probably the best way to start is to build up a picture of the research tree in which that hunch is located, and then explore it a little…

The Italian colour recipes research tree

For building up an initial view of this research tree, I began with “Original Written Sources for the History of Mediaeval Painting Techniques and Materials: A List of Published Texts” by Salvador Muñoz Viñas, pp. 114-124 of Studies in Conservation, Vol. 43, No. 2 (1998) (many thanks to Juergen W. for this!). [Though because this concentrates on published sources, there may be several significant books of colour secrets out there that it misses, e.g. MS Sloane 416.]

Now that I have pruned the (initially somewhat overgrown) research tree down to more manageable proportions, this is the view I’m currently seeing through my research window:

This research tree has four main branches I now hope to explore in more depth:

* The recipes of Johannes Alcherius (as copied by Jehan Le Begue).
– – Jehan Le Begue’s translation is in volume 1 of Mary Merrifield’s book Original Treatises.
– – Alcherius was in fact a master builder working in Milan, had access to a large number of secrets, and was still alive in the early years of the 15th century: and so would seem to be an excellent candidate for the author of the Voynich Manuscript. 🙂

* Secreti per colori, Biblioteca Universitaria di Bologna MS 2861
– – Description here, transcription here, or in Volume 2 of Merrifield.

* “Ricepte d’affare Piu Colori”
– – There’s an article on this in Archeion, Vol. XV (1933), pp. 339-347 by Daniel V. Thompson Jr, which I hope to read soon. 🙂
– – There may possibly be more in “Trial index to some unpublished sources for the history of mediaeval craftsmanship” Daniel V. Thompson Jr – Speculum / Volume 10 / Issue 04 / October 1935, pp. 410-431.

* MS Sloane 416, “The Venetian Manuscript”.
– – This manuscript also contains a (brief) description of ciphers, which makes it doubly interesting to me.
– – Parts of this might be in Dutch, but I can’t tell properly from the description.
– – I can’t find any good description of this. I’ll probably have to spend a day at the British Library…

Have I missed anything important? Please say if I have! Oh, and I’ve ordered a copy of Mark Clarke’s (2001) “Art Of All Colours”, which looks to be extremely interesting as well….

In a recent post, I discussed a large number of features of the Voynich Manuscript’s Quire 20, to try to get under its vellum skin (so to speak). One thing I’d add is that if you compare the vellum colour of the front (recto) and back (verso) of pages f103 through to f108…

…you can see (I think) if you constrast enhance these…

…that f104, f105, f107 and f108 appear to match each other quite well.

Hence – given that I think f103 may well be separate anyway – I infer from this that f104, f105, f107 and f108 may well have been folded and cut from a single piece of vellum: and hence that they might very well have sat next to each other in Q20’s alpha state. Hence f106 and the missing f109-f110 bifolio may well have been from a different piece of vellum, and so (given that I suspect f105r was the first page of the book hidden in the quire) may well together have formed the two most central bifolios of the quire.

So: given all the above, and that there seems to be a good chance that f108v and f104r originally sat next to each other (as discussed before), I suspect we now know enough to reduce the large list of bifolio permutations for Q20’s original state down to just four good candidates (I’ve only listed the first folio of each bifolio pair for convenience):

a) f103 : f105 f108 f104 f107 f106 [f109]
b) f103 : f105 f108 f104 f107 [f109] f106
c) f103 : f105 f107 f108 f104 f106 [f109]
d) f103 : f105 f107 f108 f104 [f109] f106

Of course, I may be wrong… but I do now think there’s a high chance that this is basically correct.

But to use the block-paradigm trick (i.e. to decrypt a cipher by finding a separate copy of the text from which it came) with these possible candidates, though, we need to also find a structurally matching copy of the hidden book’s plaintext.

So the big question is surely this: how on earth do we find a copy of Quire 20’s plaintext?

Johannes Alcherius

Elsewhere in Cipher Mysteries, I mentioned BNF MS Latin 6741, which is a collection of 359 recipes collected together in Paris by Jean le Bègue / Jehan le Bègue in 1431, and which was discussed by Mary Merrifield in her 1849 book Original Treatises, Dating from the XIIth to the XVIIIth Centuries, on the Arts of Painting.

Yet having now read in rather more depth about Jean le Bègue’s recipe collection here, I suspect that the answer lies not in Paris (the city which BNF MS Latin 6741 seems never to have left) but instead in Milan, and specifically with Johannes Alcherius – arguably the key author whose recipes Jean le Bègue ripped off collected together.

However, the big problem is that while Alcherius wrote in Italian, Jean le Bègue wrote in Latin – so what we have in BNF MS Latin 6741 is actually a Latin translation of Alcherius’ Italian original.

So if I’m right about this, it would mean – unfortunately – that rather than finding the Italian plaintext from which I believe a large part of Quire 20 was derived, I strongly suspect that BNF MS Latin 6741 instead contains Jean le Bègue’s Latin translation of that same Italian plaintext.

So… even though I suspect that the block paradigm trick may have got me close to the finishing line in this instance, the Voynich Manuscript’s secrets still continue to elude us. Close, but no cigar. Or pizza. Oh well! 🙁

How to cross the Quire 20 line?

It seems to me that the single most important piece missing from this jigsaw is the Italian plaintext of Johannes Alcherius’ recipes for colours. I can see two possible routes to achieve this:

(1) Reverse translate Jean le Bègue’s Latin back into the Italian plaintext from which it was derived. Which would be exquisitely nuanced, and very hard to get right, but just about possible all the same. Or…

(2) Find fragments of Alcherius’ recipes floating in other documents, but in their original Italian form (rather than in Jean le Bègue’s Latin translation). It may be that, somewhere in the far recesses of Academe, someone has already searched for this, perhaps as part of a compeletely separate study. But if so, my own digging has been utterly unable to find it.

Perhaps you will have better luck, though!

Things you can help with

Does someone have (or can get access to) a PDF copy of “The recipe collection of Johannes Alcherius and the painting materials used in manuscript illumination in France and Northern Italy, c. 1380-1420” (1998) by Nancy Turner that they can send me, before I start throwing my money at Maney Online (now part of Taylor & Francis)?

Or a copy of “Painting Techniques : History, Materials and Studio Techniques, Proceeding of the IIC Dublin Congress, 7-11 December 1998”, where the same thing also seems to appear?

Or… does anyone have a study listing all pre-1500 Italian colour recipe manuscript fragments? (For what it’s worth, I was unable to find any mention of Alcherius or Jean le Bègue in Thorndike.)

Back in August 2010, I posted up some observations on the Voynich Manuscript’s Quire 20, which included:
* Tim Tattrie pointed out that ‘x’ appears on every folio of Q20 except the first (f103) and the last (f116)
* I noted that these ‘x’ characters often sat next to ‘ar’ and ‘or’ pairs, e.g. arxor / salxor / kedarxy / oxorshey / oxar / shoxar / lxorxoiin, etc.
* Tim Tattrie also pointed out that the paragraph stars on f103 and f116 are notable because they don’t seem to have tails
* The tail-less paragraph stars on f103r looked to me as though they had been added in at a later stage
* Elmar Vogt pointed out that most of the paragraph stars followed an empty-full-empty-full pattern, except for “f103r, f104r and f108r”
* “The notion that Q20 originally contained seven nested quires (as per the folio numbering) seems slightly over-the-top to me”
* f103r doesn’t “look” it should be the first page of Q20, but f105r (with a nice ornate gallows) does:-

From these, I tentatively concluded (way back then) that Q20 might well therefore have originally been written as two separate codicological parts, which I proposed calling “Q20A” and “Q20B”. It was certainly an interesting suggestion… and, as I’ll explain below, one that I suspect was quite close to the truth, though admittedly not the whole story.

The “ytem” / “ydem” star tails

I then posted again in September 2010, where I proposed that the tails of the paragraph stars had been written first: and that they had been then been accessorized with a star to hide them from view.

That is, the tail is the meaning (they all read ‘y’ if you look closely), and the star is the deception. But what does the ‘y’ mean? Well, lists in countless medieval documents are very often itemized using the word “item” or “ytem”: which is why we use the word “itemized”, of course. So it seemed (and still seems) overwhelmingly likely to me that the paragraph star tail ‘y’ was short for “ytem”, and that each Q20 star with a tail marked the start of an item.

If you count up all the paragraph stars with tails, you get the following numbers:

f103r 0 + 19 (i.e. no stars with tails, but 19 stars without tails)
f103v 0 + 14
f104r 13
f104v 13
f105r 10
f105v 10
f106r 15
f106v 14
f107r 15
f107v 15
f108r 16
f108v 16
f109r ?? \
f109v ?? -\ missing
f110r ?? -/ bifolio
f110v ?? /
f111r 17
f111v 5 + 14 (something a little odd would seem to be going on here)
f112r 12
f112v 12 + 1
f113r 16
f113v 15
f114r 13
f114v 12
f115r 12 + 1
f115v 13
f116r 0 + 10

Add up all the stars with tails and you get 264: if the four pages from the missing central bifolio (f109 / f110) each contained 15 stars, we would seem to get a total of closer to 264 + 4 x 15 = 324 items (as opposed to starred paragraphs). I don’t know exactly what that means, but it is what it is.

Quire 20 “Block Paradigm”

When I revisited Q20 in 2014 , it was to propose that we might also profitably look for a “block paradigm” match to Quire 20. By this, I meant that we should go a-hunting for an existing book of secrets from the right time frame from which Q20’s contents might well have been copied / encrypted. As a pretty good candidate, I suggested BnP MS. 6741, which contains a set of 359 numbered recipes (plus various rhymes) compiled from various sources by Jean le Bègue / Jehan le Bègue [1368-1457] in Paris in 1431: in Latin.

But then again, when in the past others have suggested that this section might just as well contain 360 elements (as in per-degree astrology), or even 365 elements (one for each day of the year), it has been pointed out by way of response that the number of starred paragraphs doesn’t seem to fit: we have too many stars. “My God, it’s (too) full of stars”, one might reasonably say (if you are a cinema buff, that is).

This is because if you take Q20 as a whole, you would expect its total number of paragraph stars to be around 323 + (15 x 4) = 383, which is about 20-25 stars too many for (what I, at least, consider) the most likely scenarios (359 / 360 / 365).

Yet if you restrict yourself to “ytem stars” (i.e. stars with tails), it seems that you end up with roughly 324 “ytems”, which would seem to be 35 or so stars too few.

So how do all these odd-shaped jigsaw pieces slot together? Quire 20 would seem to be quite the three pipe problem, as Sherlock Holmes would have (fictionally) said. 🙂

…or is it?

How to read this hidden book

Building on all of the preceding observations (and inspired by comments recently left here by Rene Zandbergen), here’s how I think Q20 was written, and how we should try to “read it” – that is, how our eyes should sequence its pages and comprehend its content.

It seems likely to me that the ornate gallows character on f105r marks not the start of a quire, but the start of a book – or, at the very least, the very first “ytem” in that book. Furthermore, if we group all the pages with “ytem” tailed paragraph stars together and put this at their front, f105r would have been either (a) the first part of a free-standing book (“Q20A”, as I proposed before), or (b) the first part of a book whose presence inside Q20 was concealed by fake paragraph stars. Either way, I now feel confident that we should be reading f105r as if it were the first part of the hidden book.

In which case, the fake-looking paragraph stars on f103r, f103v and f116r would indeed be fake, added to visually conceal the start and end of the book within Q20. f103r has nineteen of these fake stars crammed down its left margin: the more you look at them, the more fake they look, I think.

So: if we put the f103-f116 “fake paragraph star” bifolio to one side, and place the f105-f114 bifolio as the outermost ‘wrapper’ of the hidden book, the question then comes whether we can infer from any other statistical or visual properties what the nesting order (and orientation) of the other five bifolios inside it was (i.e. when the book was in its original, ‘alpha’ state).

Tim Tattrie insightfully noted in a comment here that:

* “lo” as a separate word is only found in f104r, f106r and f108v.
* “rl” as a separate word, or word beginning is only found in f104r, f108v and f113r.
* “llo” as a series of letters is only found in f104r, f108v, f111v, f113v, f116r

I think this suggests some kind of semantic, content-based link between f104r and f108v: and hence that f108v and f104r may well have originally sat facing each other in the original bifolio layout.

Unfortunately, I haven’t yet noticed any Q20 colour transfers dating back to its earliest phases of composition. The big reddish smudge on the top right of f103r (where some of the Voynichese letters inside it have been badly emended in a later hand) seems to have happened after the pages were in their final nesting order, and the same seems true of the (initially green, but then fading away quickly) stain at the centre top of the Q20 pages.

There are also some tiny green paint splodges on the very right hand edge of f104r: but these seem to me to have almost certainly been added by accident when the heavy painter was painting Quire 19 (f103 is a little bit smaller than f104, so the outermost edge of f104 stuck out a little beyond f103, catching the drips from the paint brush).

All in all, the problem with trying to reconstruct Q20’s original nesting order is that it was left in a very clean (i.e. unmessed-around-with) state, probably because it was the least visually intriguing quire. So unless we can find a magical Raman-style way of picking up close-to-invisible inter-folio paint transfers from the paragraph stars, we don’t seem to have much else to work from.

Hence at this point, I’m basically out of ideas: are there any other sources of information (whether visual, statistical etc) you can suggest that might help us reconstruct Quire 20’s original folio sequence?

You might think I’d be pleased by the appearance of another Voynich statistics study (Voynich Manuscript: word vectors and t-SNE visualization of some patterns), courtesy of those well-known peer-reviewed online journals Reddit and Hacker News. [*] After all, statistical experiments are – if carefully planned and executed – beyond all reproach, surely?

But there is a big problem (arguably a meta-problem) with this: and it’s one that’s been around for a very long time.

Even back in 1962, Elizebeth Friedman – having been a top US Government code-breaker for several decades – was able to note that all attempts to decrypt the Voynich Manuscript as if it were a simple language or single-substitution alphabet were “doomed to utter frustration”. That is, if you wind the clock back half a century from the present day, it was already clear then that Voynichese’s curious lack of flatness was strongly incompatible with:
* natural languages
* exotic languages
* lost languages
* monoalphabetic (simple) substitution ciphers, and even
* straightforward hoaxes

Unfortunately, the primary assumption of flatness is precisely the starting point of a large number of statistical studies carried out on the Voynichese text ever since.

Why Is Voynichese Not Flat?

A long succession of (actually pretty good) past statistical studies has revealed that Voynichese has an abundance of mechanisms that give it internal structure, not only in terms of letter adjacency and within words generally, but also within lines, paragraphs, and pages. Yet while all natural languages do work to plenty of orthographic rules, none of them (from this far back in time, at least) has orthographic conventions that extend so far into the high-level page layout.

In Voynichese, you can see these “supra-orthographic structures” in such places as:
* Horizontal Neal sequences (stereotypically manifesting themselves as pairs of single-leg gallows placed about two-thirds of the way along the topmost line of a paragraph or page
* Vertical Neal sequences (the first letter of each of a series of adjacent lines, forming a putative column of letters, and very probably distorting the agrregate statistics for the first character of each line)
* Vertical free-standing key-like sequences
* Substantial difference in word structure within “labels” (short pieces of free-floating text, typically inside or beside drawn features)
* Grove “titles” (small fragments of right-justified text tagged onto the end of paragraphs, e.g. on f1r)
* Small text_size:dictionary_size ratio
* Multiple repetitions of high frequency words (daiin daiin, qotedy qotedy, etc), etc

[Just about the only supra-word-level orthographic structure we can directly match is the change in frequency stats for the last letter of a line. In natural languages, we often see a hyphen placed there, while in Voynichese we often see EVA ‘m’ or ‘am’: so I would be unsurprised if these are essentially the same thing.]

Each of these features (which I’ve discussed in more detail elsewhere on this site) on its own would be annoying enough to account for, if (say) you were trying to reconcile Voynichese with a conventional language. However, put them all together and you suddenly get a glimpse of what we’re really dealing with here: something arbitrary, painfully complex, and extremely unlanguage-like.

If, as per almost all natural languages and ciphertexts, Voynichese did not have these features, we would happily describe it as “flat”, and it would be utterly fair and reasonable for people to throw their home-grown statistical toolkits at it in the reasonable expectation that something might just emerge from the process.

However, Voynichese is not flat: and so this kind of simple-minded approach is 99.9% certain to reveal nothing of any genuine novelty or insight. Sorry, but that’s just the way it is.

So, What’s The Answer, Nick?

If you want to do statistical analysis on the Voynich Manuscript that genuinely stands a chance of producing insightful and helpful results, you really need to put the Voynichese text through some kind of normalization filter before analysing it: by which I mean you need to condition the worst parts out.

The best starting point is to restrict your scope to one of the two large relatively homogeneous blocks of text:
* Quire 13 (but without labels, and without vertical sequences) – though note there is a long-unresolved suggestion that Q13 may have originally been composed in two parts / phases, not coincident with the final binding order
* Quire 20 (but without f116v) – though note there is also a long-unresolved suggestion that Q20 may have originally been composed in two parts / phases, and also not coincident with the final binding order.

Doing this should sidestep the thorny issues (a) of Currier A vs Currier B, (b) of text vs labels, and (c) of space transposition ciphers (because I don’t recall Q13 and Q20 having and “oro ror”-like sequences). [Personally, Q20 would be my preferred starting point.]

I would also strongly advise filtering out any matched pairs of single-leg gallows that fall on any single line, along with the (usually shortish) text sequence that sits between them: and any ornate gallows too.

All of which leaves the tricky issue of how best to normalize page-initial, paragraph-initial, and line-initial letters. The jury is still well and truly out on these: which probably means that evaluating them would be a good use of statistical analysis. Which also probably means that nobody is going to actually do it. 🙁

Finally: once you have got that far, all you’re left with is… the truly humungous issue of how best to parse Voynichese. Is EVA ‘ckh’ one letter, two letters, or three? Should EVA ‘qa-‘ and ‘qe-‘ always be interpreted as if they are copying errors for EVA ‘qo-‘? Should each of EVA ‘or’ / ‘ol’ / ‘ar’ / ‘al’ be read as a pair of letters or a single (tricky) verbose cipher glyph? Does ‘ok’ encipher a different token to ‘k’? Is ‘yk’ two letters or one composite one? And so forth… the list goes on (and it’s a very long list).

But unless you can find a way to see clearly past Voynichese’s supra-orthography, you’ll probably never get even remotely close to anything that interesting with your own Voynich statistics. Just so you know! 😐

[*] Tongue planted firmly and immovably in cheek.

I’ve been persuaded by the lovely people at the London Fortean Society to give a talk next month (25th February 2016, Bell pub in Petticoat Lane The Pipeline, 94 Middlesex Street, London E1 7EZ, 7.30pm for an 8pm start, £4/£2 concs) on the weird (and occasionally wonderful) Voynich Manuscript.

If you haven’t been to an LFS event before, they start about 8pm with a “Fortmanteau” (a Fortean news round-up), followed by the main speaker for most of an hour. Then, after a 20-minute break, there’s a Q&A, finishing at 10-ish, optionally followed by a drink and a chat at the bar. As normal, I’m expecting to be assailed with questions on just about every cipher mystery going: which should be excellent fun. If any Cipher Mysteries readers plan to come along, please let me know!

If you don’t already know about Charles Fort, then shame on you! (Only kidding!) Fort liked to collect reports of phenomena that the science of his day couldn’t account for, which he edited and published in 1919 as The Book of the Damned: as a result, “Fortean” has become a useful adjective to hang onto anomalous data which sit uncomfortably with the so-called wisdom of the day. Hence “The Fortean Times”.

Is the Voynich Manuscript Fortean? For many people, it is: they would argue that scientific and historical investigations have so far revealed little of genuine interest or certainty, and that all the while it remains unreadable / uncrackable it is an anomalous artefact.

Yet for historians, this “Fortean Voynich” notion is perhaps something of a misdirection: there are plenty of old objects the smartest historians out there can as yet make no sense of – but does that necessarily mean that they are anomalous? Instead, might they simply be under-studied?

For the moment, though, perhaps there is truth enough in both camps: and that, as Charles Fort said, there is “…nothing, in religion, science, or philosophy, that is more than the proper thing to wear, for a while“. 🙂

I hope to see some of you there: and here’s the blurb I put forward for the LFS’s next flyer:

The Blurb

The 500-year-old Voynich Manuscript is renowned as “the world’s most mysterious manuscript”, as well as “the Everest of historical code-breaking”. With 200+ pages of unfathomable text, strange circular diagrams, and numerous drawings of impossible plants and tiny naked women, some consider it beguiling: others think it totally mad.

But the harder we strive to find explanations for the Voynich’s countless oddities, it seems the less we know. Is it insanely brilliant, or brilliantly insane? Even after a century of study, nobody can be sure. The only category it truly fits is Charles Fort’s “Damned Data” – phenomena for which science cannot comfortably account.

As a result, there is a plethora of Voynich theories, across a reassuringly Fortean panorama that ranges from conspiracies to lost South American civilizations to time-travelling aliens. Might one of them be right? Or are we doomed to spin round in circles, forever unable to make sense of this most intellectually cursed of artefacts?

Nick Pelling, a Voynich Manuscript researcher for more than a decade, will guide us through its wobbly history, unknown science and mad theories, and will happily answer any question we have on unsolved historical ciphers.

Back in 2012, I got (briefly) excited by the hypothesis that the marginalia on f116v of the Voynich Manuscript might well have been added in the library of a monastery not too far from Lake Constance, inbetween Switzerland and Southern Germany (and not too far from Rudolf II’s Imperial Court at Prague, where the manuscript appears to have ended up).

And then a few days later I got excited all over again by the follow-on hypothesis that this Swiss library may have been part of a Franciscan monastery. If the “bearer” who brought the Voynich Manuscript to Rudolf’s court (and to whom Rudolf paid the wondrous sum of 400 ducats) was himself/herself a Franciscan friar/nun, that might help explain its attribution to Franciscan monk Roger Bacon.

It’s a plausible story, sure, though not necessarily a highly probable one for the moment. But all the same, this might possibly give us a good idea for a brand new kind of haystack to rake through…

Franciscan Monasteries in Switzerland

St. Francis famously exhorted his followers to study in ways whereby “the spirit of prayer and devotion was not extinguished”: which makes it likely that just about every Franciscan monastery and friary we could consider would contain a library of some sort.

Indeed, some Swiss Franciscan monasteries had very famous libraries: Schaffhausen had a chained library (“Kettenbibliothek”, if you want to search for “Kettenbuch” in German). Here’s what the Biblioteca Medicea Laurenziana later looked like (a little later) with all its chained books:

Chaining books actually freed them, by making them available to more people to study: so it’s entirely possible that the Voynich Manuscript had a chained wooden cover for part of its pre-Rudolfine life. Here’s what an individual Kettenbuch from 1484 looks like:-

The Schaffhausen Ministerialbibliothek was (if I translate the nice German account of it here correctly) formed in 1540, manuscripts mainly from the Benedictine Allerheiligen (All Saints) monastery library, but also “eight manuscripts and six incunabula” from the Franciscan chained library (formed in 1509). Books (such as Erasmus’ Omnia Opera) were added from 1540 onwards.

How do we know this? Because of a library catalogue (“Chronik der Stadt und Landschaft Schaffhausen”) prepared by Johann Jakob Rüeger (1548-1606) in 1589, then updated in 1596, and apparently printed in 1884-1892 (it seems to have been partially converted into a database on ancestry.com, but I’m don’t have a subscription to that). You can see many individual pages in the extract of a modern book here (with a fair bit on the Schaffhausen Franciscan library on pp.45-47).

Here are some other Swiss Franciscan monasteries that had libraries:

* Fribourg Monastery. According to this page (with links to 13 digital copies of mss from there):-

The library contains about 35,000 volumes, 10,000 of which date from before 1900. The majority of the books can be accessed via a card catalog. The old library can be traced back to Guardian Friedrich von Amberg; 18 of his volumes have been preserved. During the monastery’s golden age in the 15th century, the superiors collected mainly sermon and study literature. The Franciscan Monastery was able to preserve its library on site; it contains 80 medieval and 100 post-medieval volumes of manuscripts (not catalogued), as well as 136 incunabula and 80 post-incunabula.

* Lindau island had a convent of the Third Order of St Francis: this survived the Protestant Reformation by converting to Protestantism.
* Konstanz
* Bellinzona
* Bremgarten (Aargau)
* Königfelden Abbey
* Wesemlin, Lucerne (has the Provinzarchiv der Schweizer Kapuziner, though presumably this was slightly later?)

…and doubtless a fair few others besides.

Clearly, this looks like it could be a substantial set of haystacks to be going through to find a single Voynichian needle. Is there anything out there that can help us?

A Swiss Needle Magnet?

It seems that there might be, in the form of the three-volume Handbuch der Historischen Buchbestände in der Schweitz that lists numerous ancient Swiss libraries, many of which have descriptions of historic catalogues of those libraries.

* Volume #1: Aargau Canton to Jura Canton
* Volume #2: “>Lucerne Canton to Thurgau Canton
* Volume #3: Uri Canton to Zürich Canton

Unfortunately, only volume #2 of this is currently online (I think, but please correct me if I’m wrong!); and many collections that might reasonably be listed are (according to the German Wikipedia page) absent. Moreover, lots of the interesting stuff is in journals such as Helvetia Franciscana that are not currently online, e.g.

* Schweizer, Christian: Kapuziner-Bibliotheken in der Deutschschweiz und Romandie–Bibliothekslandschaften eines Reform-Bettelordens seit dem 16. Jahrhundert in der Schweiz nördlich der Alpen. In: Helvetia Franciscana 30/1 (2001), S.63
* Mayer, Beda: Der Grundstock der Bibliothek des Klosters Wesemlin. In: Helvetia Franciscana 7 (1958), S.189
* Mayer, Bea: Kapuzinerkloster Freiburg, In: Die Kapuzinerklöster Vorderösterreichs. In: Helvetia Franciscana 12, 7. Heft (1976), S. 207-216.

…along with other journals such as Librarium which (thankfully) have been placed online, e.g.

* Kronenberger, Hildegard: Das Kapuzinerkloster Wesemlin in Luzern und seine Bibliothek. In: Librarium 9 (1966), S.2

And the bigger problem is this: because the Voynich Manuscript had without much doubt left its (probably monastic) library by (say) 1613 or so, what we actually would like is a list of pre-1613 Swiss Franciscan monastic inventories to have a look at, based on the small (but likely non-zero) likelihood that one of them might well list a reference to a book resembling the Voynich Manuscript. Yet this was (I think) not at all the challenge the Handbuch der Historischen Buchbestände in der Schweitz was set up to meet at all.

But… are there any of those old inventories from Franciscan monasteries still in existence all? Personally, my head’s still spinning from trying to take in all this stuff, to the point that I’m still a very long way from being able to tell. But perhaps Cipher Mysteries readers will fare better than me (even one would be nice)… good luck!

Contrary to what some Voynich people like to assert, the point of History is really not to allow a thousand speculative flowers to bloom. Rather, the idea is to work so closely with the available evidence that we can cull bad theories inconsistent with it. So for the Voynich Manuscript, where pseudohistorical theories run amok (typically blindfolded, and with a Japanese carving knife in each hand), the Voynich battle is only just beginning.

What I’m trying to say is this: because History is primarily the study of what actually happened, it is by nature a cruelly eliminative mistress – its heart is one of disproof, and that’s OK. Yet in the case of the Voynich Manuscript, the ongoing absence of incontestable evidence (let alone “smoking gun” proofs) has meant that her machete has rarely been employed in the Voynichian garden. By this measure, I’d say that there has been little done to date that would be worthy of the title “Voynich History” (understanding its post-Rudolfine provenance has, though often fascinating, so far been more of an exercise in historiography than in actual History).

Sorry, people, but this is going to have to change.

Radiocarbon Dating? *shrug*

You might reasonably have thought that the 2009 University of Arizona radiocarbon dating of the Voynich Manuscript’s vellum (to 1404-1438 with 95% statistical confidence) should surely have brought some kind of veridicality to the whole arena. Yet it has apparently achieved nothing: Rich SantaColoma merely shrugs forwards (asserting that if the vellum was used one or more centuries after it was manufactured, who cares? So what?), while Diane O’Donovan merely shrugs backwards (asserting that the date of its actual writing-down-ness is meaningless; and given that the origins of its contents ‘clearly’ lie centuries if not millennia earlier, who cares? So what?).

These two are good examples of the rampant denialism that currently passes for normality in the Voynich world. So, have any Voynich theories been modified as a result of the radiocarbon dating? My guess is that so far no, probably not even one has.

As for me: even though I have a technical criticism of one of the test’s four samples (the apparently-earliest radiocarbon dating sample was taken from one of the most contaminated parts of the manuscript, I strongly suspect affecting the very consistent results yielded by the other three), I’m far from a critic of radiocarbon dating. If we broaden slightly to the wider dating range implied by the three most consistent samples alone, that range does sit comfortably with other solid dating evidences I have pointed to over the last decade.

So I don’t see the radiocarbon dating as inherently problematic (beyond a technical quibble over its claimed precision). Yet it has made not a ripple: why?

“I Am Your Father”? Riiiight…

In the mixed up world of Voynich research, even though there is no “mainstream” central account of the Voynich Manuscript and its 15th century origins, people such as Rich SantaColoma and Diane O’Donovan invest much of their efforts into kicking back against figures they perceive as representing that mainstream (usually either me or Rene Zandbergen, whose fact-centred Voynich website I have nothing but praise for).

A lot of the time this comes across as a kind of rage against the Evil Empire: as if Rene Zandbergen and I are Darth Sidious and Darth Vader, darkly paternalistic forces looming over them all, conspiring against their Rebel Alliance’s plucky fight for historical freedom of expression. I guess this would also paint Rich and Diane as Luke Skywalker and Princess Leia, twin mystery researchers mysteriously separated at birth, but fighting on the same side: I doubt I’m even close to being the first person to note this.

But really: me? I’m certainly not their father, and I’m definitely not the mainstream. All I’ve done for the last fifteen-odd years is to try to date tiny features of the Voynich Manuscript by isolating them and locating them within their individual micro-histories. All of which has been worthy but boringly mundane; and about as far from constructing some kind of theory-zapping epistemological Death Star as you can get.

[Yes, I did put forward a theory about one possible author back in 2006: but if I had found other candidates that matched the dating evidence even half as well I’d have written about them too. All the dating evidence remains.]

The Voynich Mainstream

Whether or not it suits Voynich denialists, the things that actually form the Voynich mainstream are those historical dating observations that chime most solidly with the (I would suggest slightly wider-range) radiocarbon dating. This is because radiocarbon dating is almost never used in isolation: it is at its strongest when carefully combined with other kind of dating evidence. In the case of the Voynich Manuscript:

* The presence of 15th century marginalia in the zodiac section (the “zodiac hand”) suggests a latest date of 1500
* The presence of 15th century number forms in the quire numbers suggests a latest date of 1500
* The transitionary numbering style used for the quire numbers suggests a latest date of 1500
* The presence of parallel hatching suggests a earliest dating in Germany from 1425, in Florence from 1440, in Venice from 1450 or later: and a latest dating of about 1480 / 1490 everywhere
* The presence of Islamic-influenced geometric designs on the albarelli-like “barrels” in the zodiac section suggest a date range of 1450-1475
* The dot pattern on some of the ‘pharma’ glassware (i.e. f89r1 and f89r2) is strongly reminiscent of post-1450 Murano glass decoration, and (to the best of my knowledge) is found nowhere else until significantly later
* The baths in the balneo quire suggest a dating of not later than 1500, because that was when baths sharply fell from fashion in Europe because of their [incorrect, but persuasive] association with syphilis (Klebs, 1916)

So the Voynich mainstream is not so much a set of paternalistic individual theorists for the denialists to kick back against, but rather a constellation of specific ideas that point to dating ranges consistent with the radiocarbon dating. The reason I’m forever in the Voynich denialist firing line is that I happened to latch onto much of this group of ideas a number of years before the radiocarbon dating was carried out, and gave many of them an airing in my 2006 book.

Of course, based on what has happened over the last decade, nobody in Voynich Land will accept any of this from me without more evidence turning up. But regardless of whether I’m alive or dead or flying my TIE Advanced x1 through space, the ideas that make up the Voynich mainstream are now what they are.

Yet what kind of an Intellectual Historian would want to take any of this Voynich battle on board? Or has the Voynich Manuscript gone beyond the point where there is any real hope of its being reclaimed by mainstream historians as a genuinely interesting artefact?

It’s a hard thing to admit, but I think the Voynich Manuscript tells us much more about History – and specifically about historical proof – than History tells us about the Voynich Manuscript.

Even though we have accumulated so much micro-knowledge about the Voynich (by which I mean how its ‘language’ works, and even – to a certain extent – how it was made and owned), we still have almost no genuine macro-knowledge about it at all. For all the long list of suggestive details, our Voynich knowledge in toto is little more than a forensic desert (and not obviously different to that surrounding the Somerton Man), one that remains so wide that none may cross it and live.

For all Rene Zandbergen’s accumulated provenancing, for all his patient and informed historical nuancery, teasing single strands out of Kircher’s Republic of Letters and weaving them into what are little more than semi-threads, we still know essentially nothing about what Kircher knew of it, or thought of it. We don’t even know if he ever saw the dratted Manuscript, or if the pages (or copies of pages) sent to him by Baresch are still extant in some unknown Jesuit cryptographic archive somewhere. Or (if they are) whether or not they will give us even a flicker of assistance in decrypting Voynichese (based on past form, I suspect they would not).

And yet…

So far, so nothing: but here’s the rub. Even though I’ve long held as a basic research tenet the notion that historians are better equipped at disproving fallacious claims than genuinely proving things, why do you think it is that even after all this time, the Internet is still awash with Voynich claims and theories that make little more than facile, superficial sense?

Example #1: why is it that Stephen Bax’s utterly foolish and superficial nine-word theory has found itself promoted to hits #3, #4, and #5 on a Google search for “Voynich” (as of today)? Is it really the case that nobody has noticed that his attempts at reading Voynichese as a natural language fail to explain more than 99.9% of the text, which is almost the very definition of “unsystematic”? Is it really the case that nobody has noticed that you can “read” almost exactly the same number of words by interpreting Voynichese as English?

Example #2: why is that that Gordon Rugg’s shudderingly awful CompSci misreading of Voynichese as a “language” generated by Cardan grilles still has any faint lustre of validity at all? Why is it that I still – all these years on – run into people whose view of the Voynich is not only coloured by this kind of anti-historical claptrap, but also utterly delimited by its faux postmodern stupidity? When will we ever manage to draw a line under this idiocy?

Example #3: why is it that Rich SantaColoma can still get away with miscasting the Voynich Manuscript’s eponymous 20th century finder as its forger? Though he “plays the game well”, if you multiply out all the individual probabilities that he has to bracket out to keep his ball in the court of possibility, it’s hard to see how he can end up with a net likelihood of more than one in a million. All of which is good for his personal position (in that nobody has outright disproved his assertion that Wilfrid Voynich hoaxed his Roger Bacon manuscript), but lousy for the overall discourse, in that he continues to waste everyone else’s time by fighting against whatever they say wherever it happens to run counter to his blessèd sub-one-in-a-million shot.

…and need I honestly extend the same face-palm of logical despair to Guiseppe Bianchi’s recent Youtube video on the Voynich, however lamentable it may be? I sincerely hope not: because disproving it (and the legions of other disappointingly rubbish Voynich non-theories) would be a full-time job, and I already have three of those vying for my as-yet-uncloned time.

The poverty of disproof

Do you see the underlying pattern here? That if Voynich theorists are happy to retreat to the far unpainted corner of possibility (however dwindlingly small a scrap of floor their ill-formed theories leave them to stand on), absolute disproof becomes almost as hard as absolute proof. Moreover, such theorists are then able to take that absence of comprehensive disproof as confirmation that they were somehow on the right track all along – that the inability of others to sumo-wrestle them out of the dohyo justifies their faith in their own worthless position.

In fact, I have become utterly bored with people sending me their worthless non-theories to disprove, because I know that however I respond with, they will then conjure up a counter-example that proves that my disproof was not absolute: and so use the opportunity that yields them to cock a snook at my so-called expertise.

And so I’m left with an uncomfortable conclusion about the Voynich Manuscript and the poverty of disproof: that if it is almost impossible to disprove a mad theory all the while a given loopy theorist can keep dreaming up flimsily etheral counterexamples of possibility, then we’ve all kind of lost before we even begin. At that sad point, the entire discourse has broken down into some kind of demented two-handed game, where one side has an endless supply of imaginative jokers to place on the table.

Ultimately, we’ve now reached a net position where the core discourse about the Voynich Manuscript is so painfully broken that there is almost nothing that can be said about it that will not immediately be opposed by wobbly counterassertions moulded out of outrageously weak possibility.

Specifically, there is almost nothing I can sensibly write about the Voynich Manuscript that Stephen Bax, Gordon Rugg, Rich SantaColoma and Giuseppe Bianchi cannot immediately oppose (sometimes in the same way, or more likely in four wildly different ways), should they wish: which, to me, speaks of a kind of pervasive epistemological collapse. It is as though knowledge’s graph of usefulness plotted against time has historically peaked, and is now steadily declining: that instead of knowing more, we are losing any sense of equilibrium about the internal dynamics that make up good knowledge. The model of knowledge as a medium for slowly accumulating sensible judgments is giving way to a model of rapidly accumulating possibilities, all as bad as each other: and we are all the worse for it.

Right now, nobody seems to grasp that cipher mysteries sit at an oddly hypermodern frontier – and that if we are not careful, this could be the beginning of the end for all knowledge. But nobody seems to care.