I've been thinking of all this from a broader perspective, which may be called "spatial plot theory".
Some hints which lead to it are:
- attempts to approach the decipherment from the character or token stats perspective have been vain so far
- there are certain marks (like gallows intrusion) that certain portions of the text were written "out of order" - probably not just lines were placed out of order, but spatial blocks of text
- failure to detect narration structure of the text; I'm not aware of much research in that direction, so I rely on mine own, in respect of the Voynich stars; and the absence of traces of structure strongly suggests some kind of shuffling
Now, therefore.
Suppose the text was not written in line-after-line fashion.
Note that this does not imply that the text was not written left-to-right. For example, I can first fill two words in each line, then fill two more words in each line, then yet 2 more words in each line, et cetera. Thus the text is written left-to-right, but it is not written line-after-line or even word-after-word.
Suppose that the meaningful text is written by placing blocks of it in pre-defined spatial portions of the folio (or, maybe, this is paragraph-confined - in that case, blocks are placed in pre-defined spatial portions of the paragraph). And the way in which the text is placed (and should be read) is known in advance, or is marked by certain markers (maybe some gallows).
And then you fill in the remaining space of the folio (or of the paragraph) with meaningless filler. That filler may still be valid vords, but it is not subject to be read-in in the process of decipherment. This moves it closer to the hoax theory in general, and to auto-copying theory in particular, because the simplest way to introduce filler is either to insert some general null token (like maybe
daiin), or to reduplicate or quasi-reduplicate the preceding (or the following) vord. Thus the "auto-copying" paradigm is valid, but only for a subset of the text!
Now, this will mess all sorts of statistics.
1) the token frequency is distorted
2) the character frequency is distorted, and so are the entropy values
3) this has both the effect of shuffling (the text is not to be read simply word-after-word) and of introducing null tokens - hence, the structure of narration is undetectable
How can we trace what's going on? I think, by examining the behaviour of the baseline - especially in the cases of reduplication and quasi-reduplication.
Below I attach You are not allowed to view links.
Register or
Login to view. where I placed the baselines in a few paragraphs.
Paragraph 1, line 2. Starting from
kair, the baseline jumps up. That may be attributed to the gallows intrusion. Next, the baseline jumps once more up at
shar. Once again, probably the effect of gallows intrusion. But then one would expect that the baseline will continue on the level reached at
shar. But it does not: it returns to its previous level of
kair.
Paragraph 1, Line 5. Why does
sairy lean rightwards? Same thing, paragraph 3, line 2, only
oksho leans leftwards.
And so on, there are more examples even in this one folio. Note that f1r, by its nature, would be either the very first or the very last folio written in the process of preparing the MS. In both cases, one would expect great care and attention put into its production. And, indeed, it's much more neat and less messy in writing than many other folios. Yet... the baseline jumps.