The Voynich Ninja

@geoffreycaveney, I've been thinking more about your idea of the equivalence of [f], [p], and [d], and I think this idea has some merit regardless of what any of the glyphs might sound like or mean. In their study of Voynichese word structure, Marco and Emma talk about a couple of statistical facts that caught my attention:

• [e] or [ee] can follow [t], [k], [cth], [ckh], [cfh], and [cph]. They cannot follow un-benched [f] or [p], though.
  
• [ch] and [sh] can precede any gallows, benched or unbenched, but not a gallows that's followed by [ch] or [sh].
  
• The core of a Voynichese syllable can be any gallows, benched or unbenched, or [r], [l], [s], [ls], or [d]. Besides the gallows, the only core letter that can be preceded by [ch] or [sh] is [d].
  

So I played around on Voynichese.com, searching for [*p*] and [*f*]. As we all know, vords containing these two glyphs tend to stick to the top lines of paragraphs, and labels. I looked for examples that had >1 token, and then searched for the same vord, except with [chd] in place of [cph] or [cfh]. Lo and behold, almost all of these turned out to be real vords, with token counts in the same order of magnitude as the version with [f] or [p]. Many were found later on the same pages. I do believe you may be onto something, Geoffrey.


I feel foolish asking this, but when you guys do statistical analyses on the VMS, do you just download the Takahashi transcription to a text file, open it with a word processor, enter Boolean search strings, and record the count of search results? Statistically analyzing texts isn't something I have much experience with, but am interested in learning. I want to do a more proper analysis, where I rank all types containing [p] or [f] by token count. Then next to each one in the second column, I want to compare the token count for the corresponding type containing [d] in place of [p] or [f], or [chd] or [shd] in place of [cph] or [cfh].

I'm warming to the hypothesis that [ch/sh + gallows] = [ee/se + gallows] = benched gallows. Which means I'm also warming to the idea that [ch] = [ee] and [se] = [sh]. I'm formulating a theory by which Brian Cham and David Jackson's "curve glyphs" [c], [e], [h], and [s] all have the same value, and that every vord can have between 0 and 6 of these "C-curves". There are rules about how these C-curves are connected, based on how many there are what other glyphs the vord contains, but the point being that the number of C-curves is a fundamental property of any given vord, an independent variable with seven possible values 0~6.

I was dissatisfied with existing transcripts, both in terms of accuracy and in terms of interpretation, including the Takahashi transcription. So I created my own.

Realizing that there were numerous ways to interpret some of the glyphs (and some of the combinations), I ended up with three specialized fonts that are integrated with four transcripts (and I have subvariations on those transcripts).

I also organized them a little differently from existing transcripts, like the Takahashi version (e.g., for example, I have a version without line breaks). I wanted at least one that made it easier and quicker to search for patterns that occur at the beginnings and ends of tokens. The existing transcripts mark line ends and paragraph ends, which interferes with word breaks, which means that not all of them (in their raw form) provided a way to evaluate an unbroken line without modification or complicated search patterns, so I felt it was important to have a version that enabled this to cut down on the search-and-replace time.



I also have numerous versions that turn the characters into patterns. For example, turn gallows and straight-glyphs into patterns of pipe symbols, and curved-glyphs into patterns of cee-shapes.

I even looked for embroidery-style patterns (which, in some parts of eastern Europe and the Middle East, traditionally encode symbols) in the overall position of glyph-shapes (steganography), and also for other shapes (like maps) in the general layout of the glyphs on the page, but I'm not sure the glyphs would have to be so positionally rigid to accomplish this, so my gut feeling is that this is not the way information in the VMS (assuming there is some) is encoded.


So What is Really Going On?

To my mind, the priority-pattern in the glyphs really does hint at the possibility of a number system (which could, theoretically, be converted to something else, like letters). It's not the only possibility (it may be generated by volvelles or charts, code-blocks or who-knows-what), but I think it may be one of the more realistic ones. I think number systems deserve far more attention than they've been getting because they share these important properties with Voynichese... both are

• position-dependent,
  
• repetitive, and
  
• self-similar.
  

The Roman-numeral system was still very much in use in the early 15th century, so the CONCEPT of priority based strokes was in the head of every literate child.


Alphabetic Interpretations

[font=Tahoma, Verdana, Arial, sans-serif]The vast majority of "solutions" directly convert the glyphs to alphabetic characters, completely ignoring positional aspects and frequency distribution of letters in their language of choice. I'm in favor of computational attacks (there are some good ones) but I've noticed that even some of those come with questionable assumptions about what they are analyzing.[/font]

[font=Tahoma, Verdana, Arial, sans-serif]As for "solutions", I can't think of a single one that describes a multi-step process (I don't count two-step anagram "solutions" that are based on highly subjective manipulation of one-way cipher concepts).[/font]


[font=Tahoma, Verdana, Arial, sans-serif]Not the Only Possibility, But a Realistic One[/font]

A number system may also account for one other aspect of the VMS that has had me scratching my head for a long time. I know that there are still many people who disagree with me, but I believe the VMS glyphs are primarily based on Latin glyph shapes, with a smaller proportion derived from Greek concepts of stacking and benching. What really stands out (to me) is the way the Latin shapes not only follow Latin glyph-forms, but also position. I keep saying this over and over because it's largely ignored and I think it's crucial to understanding Voynichese. So I've asked myself a thousands times, why would someone come up with a system that follows Latin character-shapes closely enough to vaguely look like Latin without being Latin?

The best explanation I can think of so far is that if the shapes represent a numeric (or other symbolic) system, rather than an alphabetic system, then Latin-shaped endings/beginnings might misdirect people into thinking the text is comprised of letters. If so, whoever devised the system has been highly successful. Most people never look beyond letters (I saw them as letters very briefly when I first looked through the VMS, but the positional component jumped out at me fairly early and I've been trying to account for it ever since).

JKP, I've read your work on your blog and on this forum arguing that the Voynichese glyphs are indigenous to southern Europe, rather than exotic. Let me state for the record that I find your argument convincing. Many sleuths who would have wasted countless hours exploring exotic writing systems from far off places (myself included) owe you a debt of gratitude. I'm reminded of the old folktale about the man who leaves his cozy home and travels far in search of a fabled treasure, only to eventually return home and find the treasure was buried under his own stove the whole time.

Whose transcription does Voynichese.com use? I only ask because I've found it prudent to actually look at the tokens it tags when I do a search. In a not insignificant number of cases, a token it tags as a match isn't actually a match at all. Voynichese.com is an invaluable tool for dabblers like me, but not an infallible one by any means.

I think you and I are thinking along the same lines when you speculate about vords being numbers, and I talk about a classification system or non-linguistic symbolic system. My wife is an engineer and math teacher, and she suggested that vords might be thought of as algebraic terms — set patterns of variables (i.e. factors) multiplied together. Each variable in the term gets assigned a value from a limited range, and together these make up the value (symbolic meaning) of the vord as a whole.

I'll give you an example from my line of work. In obstetrics and gynecology, every patient is given a code which sums up her entire obstetric history at a glance. It's written GnPnnnn, where each n is a discrete one-digit number. The G stands for "gravida", and the number after it is the total number of times she has been pregnant. The P stands for "para", and the four numbers after it stand for total births, premature births, abortions, and live births, respectively. My wife has a very uncommon ob-gyn code, which often catches healthcare workers' attention: G1P3303. (Meaning: she has been pregnant only once, and delivered a set of live triplets prematurely.)

Back to our beloved book, it's clear that each vord is made up of component parts in a fixed pattern, and each part can be one of a short list of possibilities. I've tentatively identified the number of c-shaped curves in the body of the vord as one of these component parts, having a value that can range from 0 to 6. If I'm on the right track with this sort of thinking, then the key will be breaking down vord construction into a minimal number of independent variables / degrees of freedom, and then counting and listing all the possible values that each of these variables can have. Then it's a matter of asking ourselves: What sort of phenomena, familiar and relevant to 15th century Europeans, would this set of variables be useful in describing or classifying?

(19-11-2019, 04:52 AM)RenegadeHealer Wrote: You are not allowed to view links. Register or Login to view.Whose transcription does Voynichese.com use? 

Hi RH,
as pointed out by Nablator You are not allowed to view links. Register or Login to view., it uses Takahashi's transcription.