Re: Another explanation for dain daiin...

[Jorge Stolfi <stolfi@xxxxxxxxxxxxx>]

Hi,

After the latest flip-flop, I feel I should clarify my current views
on the VMS.

Like everybody else on this list, I feel frustrated at seeing how
little we have gained from so much work. But I think that we *did*
make progress. Here are some things that I consider satisfactorily
proved:

  * The VMS was written before 1650.
  
      Experts haven't seen anything in the physical manuscript that
      would suggest a later forgery; and Rene's discovery of Baresh's
      letter in Kircher's correspondence pretty much kills that
      theory.
  
  * The VMS is meaningful text.
  
      It has passed all the random-garbage detection tests that we
      could think of. The entropy per word looks right, and word
      frequencies follow Zipf's law. The vocabulary shows a normal mix
      of "general-purpose" and section-, page-, and paragraph-pecific
      words. Long-range statistical tests (by Perakh, Landini,
      Zandbergen, et al.) show the expected amount of "fractality" and
      section-dependent variation. Labels are mostly unique, but
      contain some "general-purpose" words. The first word on each
      herbal page looks different and is usually unique. The words
      have a complicated internal structure. And so on.
      
  * Voynichese "words" are indeed words, spaces are spaces, labels are labels.
  
      There are too many distinct "words" for them to be just letters,
      or syllables of polysillabic words. Labels have slightly
      different letter statistics and are mostly unique, but otherwise
      they have the same structure as the "words"; most of them are a
      single "word", some are two or three. Line breaks occur at
      normal "word" boundaries.  And so on.
      
  * Voynichese is not a Vigenère-style cipher.
  
      Zipf's plot for Voynichese looks like that of any natural
      language, and very unlike that of text encoded by a
      Vigenère-style cipher --- or by any system that maps one plain
      word to different code words according to its position. Also, a
      Vigenère-like cipher is incompatible with the observed word
      paradigm and constraints on character order and position within
      words --- like 1 gallows max, "y" always at end, etc.. And so
      on. The VMS "cipher" must be of a type that maps one word to one
      word, at least most of the time.
      
  * The (bi)folios were bound in the wrong order.
  
      There is a figure in the bio section that was obviously meant
      to be at the quire's center, but isn't. The "A" and "B" herbal
      bifolios seem to have been interleaved at random, and there are
      isolated herbal pages in the the middle of other sections which
      could be brought together by appropriate reordering of the bifolios.
      
  * Whoever wrote the folio and quire numbers could not read the book.
  
      It follows from the above.
      
  * The VMS is in a single language.
  
      The so-called "Currier languages" share the same word
      structure, with very similar component frequencies. They share
      many of the most frequent words; the vocabulary differences can
      be explained by change of subject or style, or by being written
      at different times. There are similar differences between all
      sections, and smaller but similar differences between pages of
      the same section. The differences are much smaller and localized
      than, say, Italian vs. Latin. They also do not seem to be due to 
      a change in spelling/encoding rules.
      
  * The VMS was written by a single person.
  
      The only writing expert opinion we have is very definite about
      that. The differences pointed out by Currier are within the
      normal range of variation for a single person.
  
  * The VMS author was not Roger Bacon.
  
      All Bacon experts say so. Raphael's comment to Marci was
      probably a guess; and even if what he said was true, Rudolf was
      surely mistaken. Marci himself was careful to say that he was
      withholding his judgement on the matter.
      
  * John Dee never owned it.
  
      Dee's detailed diaries don't mention the VMS nor its sale to
      Rudolf. It seems that John Dee met Rudolf only once or twice
      during his whole stay in Bohemia, and was clearly a persona
      non grata at the Court. The alleged identification of Dee's hand
      in the folio numbers has been substantially discredited by
      Prinke. Recall that Voynich brought Dee into the picture as the
      person most likely to have brought a Bacon original to Prague;
      but if the VMS is not Bacon's, that connection vanishes, and Dee
      is at best one possibility among thousands.
      
  * The VMS is a product of the European culture.
  
      That follows not so much from dresses and hairstyles -- which
      may have been added by an European copist/illustrator -- but
      from the general "look and feel" of the script -- including
      writing direction and page layout -- and the use of about 25
      different symbols. 
      
      But note that "product of European culture" does *not* mean the
      author was European, that the book was composed in Europe, that
      it is in some European language, or that it contains European
      subject matter. That label includes, for instance, those books
      written in America by American indians, in their native
      languages, in the 1500's and 1600's --- such as the original
      Popol Vuh.
      
  * The "michiton" text on f116v was written by the VMS author.
  
      The text does not look at all like a cryptographer's worksheet,
      as commonly assumed. The Voynichese words <oror sheey> fit
      nicely the Voynichese word paradigm, but are very rare in the
      text hence unlikely to have been copied from there; and they are
      embedded in the paragraph as if they were part of the sentence.
      Moreover, the "telephone doodles" at the top left corner 
      of f116v are in the same style as the other illustrations.

      
Some commonly assumed "facts" that I am very skeptical about:


  * Did Rudolf ever own the VMS?
  
      If Bacon is not the VMS author, I see no rason to trust
      Raphael's comments quoted in Marci's letter. He probably never
      saw the "Bacon" book that Rudolf supposedly bought; most likely
      he only heard the story -- "Rudolf's lost Bacon" must have been a
      subject of speculation for Prague scholars of Marci's time --
      and just guessed that Baresch's book was "it".
      
      Note that in Baresch's letter to Kircher there is no mention of
      Bacon or Rudolf.  In fact, Baresch offered a very different
      theory about the book's origins, and he seemed to think of it as a
      mere curiosity that was "taking up space uselessly in his
      bookshelf". That's is a strange way to describe a book that an
      Emperor once paid 600 ducats for.
        
  * Did Jacobus de Tepenecz ever own the VMS?
  
      The faded "signature" on page f1r looks to me more like a
      scribble by a librarian or late owner -- Baresch, Marci,
      Kircher, Beckx, etc.. Note that neither Baresch nor Marci
      mention Jacobus or his signature.  In fact, the "signature"
      could have been added by Voynich, in order to strengthen
      the Bacon connection.
      
      Note also that the primary sources about Rudolf's court that
      were available to Kircher and his successors would have been
      Jesuit chronicles such as Balbín's, which overplay Jacobus (a
      Jesuit alumnus and generous benefactor), and barely mention
      other "minor figures" such as Ticho Brahe, Sendivogius, Hajek,
      et al.. So it was very easy for Kircher et al. to view Jacobus
      as the most likely author/seller/owner of the VMS before Baresch
      --- just as it was for Voynich, and has been for us.
      
  * Was the Beinecke copy written by the author himself?
  
      I cannot make up my mind on this question. On one hand, some
      parts of the book seem to have been done rather hastily, yet
      there are no obvious signs of amends, insertions, marginal
      additions, etc.. There are also many things that look like
      uncorrected errors, such as violations of the general "spelling
      rules", malformed characters, apparent "mistakes" in the 4 × 17
      sequence, etc. Often it seems that the illustrator did not know
      the nature of what he was drawing, as in the "intestines" of
      f79v; and the circular diagrams are incrediby crude for a
      student of astronomy. 
      
      On the other hand, this is all circumstantial evidence; and the
      handwriting is always readable, fluent and adaptable to the
      phisical context, as if the scribe understood what he/she was
      writing.


Now some things that I consider very likely, but still not proved:


  * The manuscript was written between 1400 and 1650.
  
      That seems to be the consensus of all experts. Many assume a
      narrower range, such as 1480-1520, but the evidence offered for
      that --- letter shapes and dress styles --- is rather weak.
      "Rounded letter shapes" is an idea that could have occurred to
      anyone at any time; and dress styles may just tell us when the
      Beinecke copy was made, not when the original was written.
      Besides, the illustrator may have intentionally imitated the
      style of older books, or those may be stereotypical drawings
      that he/she had learned as a child. As for the so-called
      "sunflower", the match is far from perfect, and it could be some
      other species with composite flowers -- which are common in the
      Old World too.
      
      In any case, the date range above applies specifically to the
      Beinecke VMS, and leaves open the possibility of it being a copy
      of an older original.
      
  * The illustrations are connected to the text.
  
      The VMS would not attract 1/10 as much attention if it did not
      have all those pictures of flolickin' ladies and weird-looking
      cosmological diagrams.  If the author's intent was to camouflage
      a dangerous text, he could hardly have done worse.
      
      On the other hand, if he wanted to make the book more appealing
      to some rich fool, he should have been more explicit -- show
      gold being made, diseases being cured, virility restored, etc.
  
  * The colors are not original.
  
      The colors seem to have been applied rather carelessly, with a
      quill pen rather than with a brush; often straying outside the
      inked outline, and leaving lots of bald spots. I would expect
      the author to show more respect for his own work. In many
      drawings only a few items have been painted -- as a child
      would often do on a coloring book.
      
      The colors often seem to have been picked at random, or strictly
      by aesthetic criteria, or without respect for the figure. For
      instance, on f17r or thereabouts we see a plant with roundish
      many-fingered leaves, painted red, and a leafy flower, painted
      blue-green. Often the leaves, scales, etc. are colored in
      alternating or checkerboard fashion.
      
      There is a site out there which has images of the first printed
      herbal. The printing was in black-and white; someone started to
      colorize the book, but soon gave up, leaving many plants
      incomplete. I suspect that the painter was a kid, and that a
      similar fate happend to the VMS.
      
  * The manuscript is not encrypted.
  
      As argued above, the VMS "code" must generally map one plain
      word to just one code word; and the code words must fit the
      observed paradigm and order/placement constraints on the
      letters. The first requirement rules out Vigenère-like ciphers
      that operate on the text as a whole. The second one rules out
      Vigenère-like schemes applied to each word separately. On the
      other hand, the cipher should be historically plausible, and
      sufficiently practical to be used for a whole book.  
      
      Those constraints seem to allow only for a codebook-based scheme,
      where the code words are generated systematically by some
      mechanical process --- essentially numbers in some Roman-like
      notation. (This description could fit also an artificial language
      with "philosophical" vocabulary, like those developed by Dalgarno,
      Wilkins, and Kircher in the 1600's.) 
      
      However, the codebook theory still has some big problems.
      Codebook schemes are difficult to write and difficult to read,
      so they are typically used for short messages of vital
      importance. I still haven't heard of any VMS-size book from
      those times that was written entirely in book-code --- or in any
      kind of code, for that matter.  In fact, I still haven't 
      seen a plausible explanation for why the VMS author would have
      felt it necessary to use encryption AND a completely new 
      alphabet.
      
  * The VMS is a "transcultural" artifact.
  
      By that I mean that the book was probably written by an European
      (E) author for non-European (NE) readers, or vice-versa; or
      perhaps for NE readers by an NE author, but after European
      fashion (The Popol Vuh and native Inca chronicles would be
      examples of this last class).
      
      I think that the first two scenarios are more likely, because
      they would explain why the book contains no recognizable
      numerals, symbols or imagery, European or not: those symbols
      that the author knew would be meaningless to the intended
      readers. The second version could also explain the "michiton"
      text as an attempt by a non-E author to produce a dedication or
      colophon in what was to him a foreign language with a foreign
      alphabet.

  * The text is in some exotic monosyllabic language.
  
      If it is not encrypted, then it is plaintext in an invented
      alphabet and possibly a peculiar spelling. The word structure
      rules out European or Near-Eastern languages, but seems to fit
      those of East Asia, such as Chinese, Vietnamese, Tibetan, etc.
      
      I believe this theory is quite plausible historically; in fact I
      can think of many possible scenarios, and cite several cases
      where Europeans invented new alphabetic scripts for "exotic"
      languages. (The official Vietnamese script, by the way, is one
      of them.) This theory seems to fit all the data, and explains
      many otherwise puzzling things --- such as why 80 years of
      attempts by the best cryptographers in the world could not make
      a dent in the "code".
      
      (For the new guys: this is the "Chinese theory" that my pizza
      bet was about.)
      

    > [Gabriel:] Then "word" sizes would not be words at all and so
    > they could be phonemes or syllables as suggested in the past --
    > I can imagine Jacques and Stolfi getting ready for the pizza :-)

Well, I have already tacitly weakened my bet from "Chinese" to "any
East Asian monosyllabic language", and stretched "East Asia" as far as
Tibet. 

That's probably more than I can get away with. So, if the VMS words
turn out to be in syllables of Latin or some other "reasonable"
language, I will have lost the bet.

To those who got this far, thanks for your patience. All the bst, 

--stolfi