A new day brings a new Google Adwords campaign from Edith Sherwood (Edith, please just email me instead, it’ll get the word out far quicker), though this time not promoting another angle on her Leonardo-made-the-Voynich-Manuscript hypothesis… but rather a transposition cipher Voynichese hypothesis. Specifically, she proposes that the Voynich Manuscript may well be Italian written in a simple (i.e. ‘monoalphabetic’) substitution cipher, but also anagrammed to make it difficult to read.

Anagram ciphers have a long (though usually fairly marginal) history: Roger Bacon is widely believed to have used one to hide the recipe for gunpowder (here’s a 2002 post I made on it), though it’s not quite as clear an example as is sometimes claimed. And if you scale that up by a factor of 100, you get the arbitrary horrors of William Romaine Newbold’s anagrammed Voynich ‘decipherment’ *shudder*.

More recently, Philip Neal has wondered whether there might be some kind of letter-sorting anagram cipher at play in the VMs: but acknowledges that this suggestion does suffer from various practical problems. I also pointed out in my book that Leonardo da Vinci and Antonio Averlino (‘Filarete’) both used syllable transposition ciphers, and that in 1467 Alberti mentioned other (now lost) kinds of transposition ciphers: a recent post here discussed the history of transposition ciphers in a little more detail.

So: let’s now look at what Edith Sherwood proposes (which is, at least, a type of cryptography consistent with the VMs’ mid-Quattrocento art history dating, unlike many of the more exotic ciphering systems that have been put forward in the past), and see how far we get…

Though her starting point was the EVA letter assignments (with a few Currier glyphs thrown in), she then finessed the letter-choices slightly to fit in with the pharma plant label examples she picked: and there you have it (apart from H, J, K, Q, X, Y, Z and possibly F, which are all missing). All you’d have to do, then, is to anagram the rest of the text for yourself, sell the book rights, and retire to a sea-breezy Caribbean island.


Might Edith Sherwood be onto something with all this? No, not a hope: for example, the letter instance distribution is just plain wrong for Italian, never mind the eight or so missing letters. As with Brumbaugh’s wobbly label-driven decipherment attempts, I somehow doubt you would ever find two plausible adjacent words in the main body of the text. Also: what would a sensible Italian anagram of “qoteedy” (“volteebg”) be?

Her plants are also a little wobbly: soy beans, for example, were only introduced into Europe in the eighteenth century… “galioss” is a bit of a loose fit for galiopsi (not “galiospi”, according to “The Botanical Garden of Padua” on my bookshelf), etc.

As an aside, I rather doubt that she has managed to crack the top line of f116v: “povere leter rimon mist(e) ispero”, “Plain letter reassemble mixed inspire” (in rather crinkly Italian).

All the same, it is a positive step forward, insofar as it indicates that people are now starting to think in terms of Quattrocento dating and the likely presence of non-substitution-cipher mechanisms, both of which are key first steps without which you’ll very probably get nowhere.

31 Thoughts on “Edith Sherwood’s anagram cipher…

  1. avatar Vicky on March 26, 2009 at 7:18 pm said:

    I was fascinated by all the information Edith Sherwood got. At first, I did believed Leonardo Da Vinci wrote the VM, but couple days later I doubt that.
    If you look at page 5 of the VM, the second circle lists four sets of 17 caracters each
    The childish drawings, otherwise they could be nicer, more defined, it just couldn’t be Leo’s and the writing, well the writing looks too fine for a child

  2. Hi Vicky,

    There’s a certain kind of romantic history that seeks to join the dots between famous people and infamous things. Over the years, people have contrived ways of linking poor old Leonardo with all kinds of stuff, even the fictional Priory of Sion (*sigh*). As far as I can see, Edith Sherwood has amassed not a jot of evidence that actually links Leonardo with the Voynich Manuscript: and so she ends up investing so much time into why the basic right-handedness of the writing can be reconciled with his basic left-handedness, rather than actually accepting what the evidence is saying and starting again.

    Truth be told, I’m enough of a closet romantic that I’d love Edith’s theory to be right: but that doesn’t make it so. Oh well!

    Cheers, ….Nick Pelling….

  3. Pingback: Tweets that mention Edith Sherwood’s anagram cipher… | Cipher Mysteries -- Topsy.com

  4. “Doctor” Sherwoods analysis is faulty to the point of embarrassment.

    After viewing the claims on her webpage that the manuscript is written in “Italian Anagrams” I laughed, then about threw my laptop across the room because of how ridiculous her claim is, and here’s why:

    1) Anagrams are the rearranging of letters from a parent or subject word, to form one or more smaller words in the same language. In each example of data she ‘decoded’ the anagrams she starts with are not words in Italian, so at best they could be considered “jumbles,” which leads me to …
    2) If this is writing in a “jumble” form, there would be only as many unique tokens as there are distinct phonemes (or more likely, letters) in the medieval Italian alphabet. Conservative estimates put the alphabet in the manuscript closer to 30 than the ~20 supposedly used in the medieval equivalent.
    3) The syntactic structure of the language hasn’t been addressed. If it is Italian in root then it should be head-initial. Something to investigate.
    4) Back to the structure of anagrams: If they are anagrams, the words should also follow the same phonotactic and phonological rules of the parent language. Most notably, the “anagrams” violate rules regarding CCC (tri-consonantal clusters) and branching onsets.

    She either needs to address the material from a scientific (not using “the DaVinci Code as inspiration,) method, or, and preferable to all serious linguists, stop all together.

    BTW, if she’s going to call herself “Doctor” or add “PhD” after her name when discussing something that is clearly not a master at, I will as you all to address me as “Sir” or “Knight of the Realm.”

  5. avatar Stan Clayton on March 2, 2010 at 9:59 pm said:

    I think its very unfair to criticise Dr Eith its obvious she has puta lot of effort into her work and has the decency to share it with othersalthough she has had answers and feedback ive put a valid breakthrouugh on different forums on the beale codes and had one replys, which is worse than being criticised, STAN.

  6. avatar James Broderick II on September 29, 2010 at 2:37 pm said:

    While I thought the Da Vinci connection was unusual, being written by a child might be an angle to investigate – especially since children can make errors that might very, very easily fool a computer.

    I, personally, always suspected the manuscript to be an Alchemical text.

  7. James: while it’s entirely possible that the VMs’ scribe was a fairly young scribe (working for the encipherer), I really don’t think it was composed by someone young, not even a young Leonardo. And as for alchemy… well, just as with claims of heresy, there’s precious little in the drawings to support the idea, in fact possibly only the “green lion”-like root in the wide herbal bifolio at the back. Nobody could understand the obscure symbolism of alchemical texts even they were written in normal language, so why encipher them as well? :-p

  8. Hmm – but why not? Her theory looks not so bad. It realy may be just a XVIII century scam/spam or guy who written this may have travelled even to America’s or many countries.

    I think probably no one will use very advanced cipher to book that kind.

    Rather something very simple for him/her/friends. Just to make some sense if they known it.

    But – could someone say – why nobody checked the paint used to write it with C14 method? – maybe it will lost some of mystery if it will be just usual XIX/XX scam…

    I think if this a real coded or unusual language book the key is to in plant drawings.

    Why not even Google -> make an algorithm to check the scanned drawings and pictures of plants in books/photos that they have in they large base?

  9. wdsd: you need a good amount of material for radiocarbon dating, far more than you can get from scraping a bit of ink off a page. It would be nice if there was much we could learn from the VMs’ herbal drawings, but I think just about every medieval herbal researchers out there has had a look at the Voynich by now. We need a quite different kind of insight and/or data to move us on, not more of the same.

  10. I believe that the so called experts have failed. So, it is a positive that folks not within the circle of mainstream cryptographers are taking a fresh and “uneducated” (ie. unbiased) look at the VM. What if anything has been done with AI to try and read the VM?

  11. GEEK: it’s certainly a suggestion that I hear a lot. But actually, the experts do know an awful lot about the limits of our VMs knowledge – that is, the limits of the statistical and inferential tools we use to understand the VMs. The honest truth is that computation (whether ‘AI’, cryptological heuristics, or sheer brute force) hasn’t so far helped us a great deal, and there isn’t a high likelihood that this will reverse any time soon: I suspect the next significant steps forward will come from non-computational types of knowledge… but we shall see, I guess. :-)

  12. I read through the Wikipedia article on this, then I stumbled upon her page, then found yours.. all to ask one question lol but no one had comment sections/emails. Has anyone tried any variation of the translation backwards? I looked through a lot of the pages and.. it just looked like something I’d at least check, if I had any idea how to decipher codes..:D

  13. Julie: just as you’d expect, many people have tried this. Some examples on my big list of Voynich theories:
    * Beatrice Gwynn (who thought it was written in mirrored Middle High German)
    * Steve Ekwall (who believes that everything in the Voynich is mirrored and “simpler than you think”)
    * various Voynich-as-Hebrew theories
    The main problem, of course, is that each line of text was clearly written left-to-right: there’s even what appears to be a poem in the balneological section, which again was manifestly written with a ragged right edge. There are even some gaps left in the places you’d expect to find ornate initials at the top-left of paragraphs.

    Having said all that, even though the codicology and palaeography seem to point to a left-to-right writing order, it’s entirely conceivable that the content was arranged right-to-left to deliberately confound us. Who can say? :-)

  14. avatar Erbsenzähler on October 25, 2012 at 3:45 am said:

    The paintings are neither carefully performed
    nor in any way meaningful, but childish.
    The colors are crass.
    There are no complex arrangements
    (like shepherd with herd and dog)
    no beautiful compositions (like a bouquet),
    no reproductions of nature (like a landscape),
    no images of abstract thoughts,
    but fictitious plants, like a child would draw it.
    The drawings include simple symmetries and duplications,
    in order to fill the place of the papers
    (in exactly the way, as the humbug text fills the place, too).
    (All the leaves and blooms within a single plant are the
    reproduction of the same leave and bloom).
    (The ‘artist’ was driven by mania of symmetry,
    displayed in the little stars and in the circles).
    All the drawings seem to have no joint motive.
    What wanted to explain the author to us?!

  15. Thomas: a “mania of symmetry” is surely the sign of an ordered mind, whether or not we can see past the noise of the surface (both the pictures and the text). So, the explanation here is that we are faced with a private artefact of an ordered mind.

    The colours are indeed “crass”, though many people (myself included) have argued that a very large part (probably even a majority overall) of the colours were added a long time (i.e. a century or more) after the original composition, which would be after several reorderings and rebindings. So it would seem to be wise to avoid using the colours to make inferences about the original composition, just so you know. :-)

  16. Nick, I think you’re selling Dr. Sherwood’s theory short.

    While Dr. Sherwood supposes da Vinci to be between the ages of 8 to 10, I would adjust his age to be either 6 or 7 (maybe even 5) given the crude drawings (compared to his early attributed works), what are likely spelling mistakes or miscalculated anagrams, and his later but not distant, fully developed genius.

    I would ask you this: did you grow up with a lefty? I did… three of them, in fact. As late as the 1970’s, teachers were still forcing young left-handed students (two of my brothers, as an example) to write with their right hand. Eventually, they were allowed to change to their appropriate hand, but my mother, who went to school in the 1950’s was given no such option. Teachers then considered left-handed children as defective. Imagine the stigmas regarding left-handedness that may have existed in the 15th c.!

    The careful (yet not wholly symmetrical) hand that scribed this work is typical of any child learning their letters at school and especially one who is working with an unnatural hand preference. While no super-genius, I was able to (and did) draw as well at about 8 or 9 years of age, albeit with not as much of the imagination.

    I would highly doubt that at even such a young age da Vinci would have settled with just a simple substitution cipher and single word anagrams. Quite possibly, he linked several words or even whole sentences using special glyphs or glyph combinations as indicators. Perhaps he incorporated other languages (which he was surely learning) or possibly began experimentation with select mirror-imaging.

    This is simply the most plausible theory out there and I think it is absolutely wrong-headed of you to dismiss it.

    Btw, soybeans came from Asia with whom trade had been long practised, not the Americas, so da Vinci could very well be aware of their existence.

  17. Damon: I’ve made a special study of the cryptography that Leonardo used in his notebooks – it formed the backbone of Chapter 6 (“The Da Vinci Cipher”) of my book “The Curse of the Voynich”. And I have to say that, left- or right-handedness aside, the cipher angle of Edith Sherwood’s Leonardo theory is one of its weakest parts, because Leonardo never systematically used cryptography in anything like as sophisticated a way as we see in the Voynich Manuscript. I wouldn’t say that Edith Sherwood’s theory is the most implausible I’ve ever seen – for one, it’s basically in the right century, which is a good start at least – but innhostorical research the Devil’s in the details, and all the details I can see and check are just plain wrong. :-(

  18. Is it true that the book is written between 1408-1438 ?

    Wouldn’t that change all the accounts of theorized writers and scribes that have been discussed by Prof. Sherwood and others before her ?

    Just to confirm where are we know in the manner of encryption ?

    I have found no article dating between 2011-2012 regarding the book.

    Please advice of any related issues or updates if there are any,

    I am merely interested in a curiosity point of standing with no PHD’s or past references I can give about any of the academic studies, so forgive me if it seems pointless for me to ask.

    Thanks in advance and best wishes.

  19. Uras: samples from the vellum (the writing support material) were collectively radiocarbon dated to 1404-1438 with 95% certainty. However, from my point of view there are some minor technical issues with the sampling methodology (one of the samples was taken close to some obvious handling discoloration, which could well yield an unreliable radiocarbon date) and the resultant degree of statistical certainty when combining the samples. That is, I think the dating is broadly right, but the fuzzy date range may well be a little wider than the authors conclude. For what it’s worth, my own art history dating of the manuscript points to 1450 to 1480 as the more likely range.

    As far as I know, Edith Sherwood isn’t a Professor. She does have a PhD, but in Chemistry, not in Art History. She has also published meta-analyses of radiocarbon dating on her website that are far more critical of it than I am.

    As far as encryption goes, nobody is at all certain. Personally, I strongly suspect that the ‘4o’ composite sign enciphers ‘lo’ (i.e. ‘the’), that the ‘av / aiv / aiiv / aiiiv’ family of shapes encipher Arabic numerals, that mid-word ‘-8-‘ and end-word ‘-9’ encipher contraction and truncation respectively, and that ‘ol / al / or / al’ are all verbose cipher pairs. However, there also appear to be other confounding mechanisms in play, so proceed with caution!

  20. Zack: *sigh* Tom’s site, once again. And your point is… what, exact;y?

  21. avatar Mikey Georgeson on November 12, 2013 at 4:13 pm said:

    I am an enthusiastic fan of Edith’s thesis. If the voynich is an anonymous hoax where did the intense, paradoxical, highly creative genius that created it disappear to? My feeling is he grew up and grappled with the Enlightenment. Were there any other artists with the level of “unquenchable curiosity” needed to produce such a codex? If so who?
    I love the mapping of childhood playful curiosity into the strictures of adulthood over the move from dark age idiosyncrasy to enlightened order and objectivity.
    It makes sense that the mind that created it was in the process of being formed through playful experimentation and not in a state of fixed rationality. Edith’s idea really captures my imagination because it draws upon Leonardo’s role as a bridge between left and right brain. Bravo Edith

  22. Mikey Georgeson: I’m sorry to have to break the news to you, but the oft-repeated suggestion that Leonardo da Vinci was somehow the only Florentine (or indeed Northern Italian, or indeed European) in the whole Quattrocento smart enough to have contrived the Voynich Manuscript is piffle on a grand scale. In actual fact, 15th century Florence (and don’t get me started on Venice) was a regular hot-house for sophisticated clever-clogses, of which Leonardo is merely the one most celebrated outside of Italy (perhaps you know that Italians themselves tend to be more struck by Brunelleschi etc?)

    Alas, though Edith Sherwood’s webpages play strongly to people’s imaginations (though she must surely, over the last few years, have spent upwards of $10,000 to Google AdWords promoting her site), as far as I can see she still has no actual backing for her idea beyond hand-waving and hopefulness. In fact, her suggestion fails even the most basic codicological test (i.e. that the Voynich was manifestly written by a right-handed scribe, when Leonardo was famously left-handed), which I would find somewhat embarrassing, to say the least.

    I’m told she gets frightfully cross whenever I point out such basic shortcomings, as if it’s somehow a personal attack on her (which it plainly isn’t). As always, the truth is less conspiratorial and far more straightforward: here in the upper-left turret of Cipher Mysteries Mansions, I am besieged by earnest people proposing all manner of hypotheses about all manner of historical ciphers – but until such people deign to get around to the gritty, awkward business of finding some evidence to support their otherwise flawless theories, it all gets filed under “C” (for “cock-and-bull”).

  23. povere leter rimon mist(e) ispero.

    1. Letter(Eng) of a monk (Ital) to the ‘Mister’ of bristly hair (Sp).
    2. destitute’s letter from Rimmon may inspire (you to charity?)
    3. poor Phoenician (or pomegranate) letters about the Hesperides.
    4. A monk’s letter on the mysteries of the Hesperides.
    5. leter rimon mist(e) ispero povere.
    Purple letter on the mystery of hisperica famina

    poveri = monks: ad altro che all’onore di Dio e alla salute dell’anime, e sieno padri de’ poveri.
    2. Hispero: de pelo aspero y erizado. Saludos (answersCom)


  24. That’s the problem, isn’t it? I mean that modern cyptography assumes that at the basic level the text will turn out to be plain, pragmatic prose, in present standards of consistency, orthography and grammar with any tenses and terms devoid of variation as of allusiveness. Prose or poetry.

    Trouble is that when you look at something like Hisperica famina, or gnostic poems of the silk road etc. even if you know the script and which language it is, the translation makes little if any sense in those terms. Hisperica famina is written in Latin, but has only fairly recently bee turned into any other language, and now it has, it still makes little sense to anyone not deeply informed about early medieval Irish Chistianity. Same for the silk road hymns.

    Pictures use an easier encoding imo

  25. avatar SirHubert on November 13, 2013 at 3:37 pm said:


    The Hisperica Famina is extraordinary and there isn’t really anything else that can be described as ‘something like’ it. The portion in the Oxford Book of Mediaeval Latin Verse was translated in the notes, I think, and that was at least fifty years ago. Because so many of the words are so very obscure, getting a clean text is a major problem and I suspect that the original version is irretrievable now. But if you were to encipher the entirety of the best modern version using a standard technique of the fifteenth century or earlier, a competent cryptographer would be able to crack it today through familiar techniques such as frequency analysis and recognising common letter pairings. Which is why the Voynich, which dismisses this kind of attack with barely-concealed contempt, is so interesting.

    I am not a professional cryptographer, but I would very much doubt whether any such would make the assumptions at the basic level you describe. You have to start with analysis of the text, statistical and otherwise, and work from there.

  26. Tricia: ah, Edith Sherwood’s translation of the top line of f116v raises its scrappy little head once again. :-(

    I’d say that its lack of internal coherency, linguistic consistency, and semantic clarity along with the need to invoke additional (but unspecified) anagramming in order to explain away the aforementioned lack of coherency, consistency and clarity make this unlikely on a number of levels.

    Really, on a scale of (1) to (hatstand), I’m sure that few judges would score this particularly close to (1). Next, please!

  27. SirHubert
    I was thinking of Jane Stevenson’s translations, which are more recent. She’s also written about Theodore, who may have brought with him texts on ‘foreign’ plants about which his monks complain.

    My point wasn’t so much that the text mightn’t be technically crack-able, but that if an original text really read like Stokjo’s translation – or more recently Joachim’s – who could be sure the decryption valid, consistency and repeatability notwithstanding? I might encrypt ‘de cursu stellarum’ and you rightly decrypt it.. but we’d be no for’arder and I’m sure many voices would be raised arguing it no genuine Latin.

    To amuse myself, I once re-wrote passages from some Manichean hymns, a bit of one long poetic-technical navigational poem and a bit of Sanskrit (in literal translation) – doing nothing but rendering idiomatic forms literally or replacing a term with one that is nearly-but-not-quite synonymous. The result read as near-random and if had been the result of a decryption, I’m sure it would have been dismissed in the same way that all published Vms trans. have been so far.
    So wouldn’t you say that enciphering took off in parallel to standardised orthography because the two are interdependent?

  28. avatar SirHubert on November 15, 2013 at 4:15 pm said:


    It’s still possible to distinguish peculiar Latin from gibberish.

    Linear B is heavily truncated Greek written in an entirely unsuitable script. But it’s definitely, demonstrably Greek.

    Gibberish, even when ‘interpreted’, anagrammed, sprinkled with arbitrary vowels and otherwise mangled into a caricature of a known language, is still gibberish.

  29. In order to consider Sherwood’s hypothesis, it could be of some use to know that here lies a famous Italian anagram engine.

    Unfortunately, it is fine-tuned for modern Italian (and not medieval one), so a slightly different spelling results in failure, not counting orthographics errors.

  30. I don’t know if SirHubert is still about, but would any other person interested in this sort of thing give their opinion on whether or not the difference between the intelligible and gibberish comes down to a standardised grammar and orthography – in your opinion?

    oh crumbs – now here’s someone we should try to co-opt: the author of a work actually published, and whose title is,

    Poems in Isthmian Proto-Zoque, Manichaean Syriac, Paleo-Elamite, and Early Gandhari in Kharoshthi

    and the name of the hero is
    Daniel Deleanu

Leave a Reply

Your email address will not be published. Required fields are marked *

Post Navigation