The Voynich Ninja

Hi Emma,
I agree that working on the illustration/scribe intersection could be the best option at the moment. I am not sure I would exclude those pages from Scribe1-Herbal pages, after all they are Herbal pages and they are by Scribe1.
It is possible that also the other larger blocks could be further segmented on the basis of bigrams (and maybe other measurable features like MATTR). In particular, I read on Nick's blog that Glen Claston noted that Quire13 is made of two kinds of bifolios (Q13a "tubes" and Q13b "pools"). I wonder if the different illustrations are paralleled by language differences?

But working on scribe+illustration seems like an objective approach and, as you say, it leaves us those three parts that are large enough for meaningful analyses.

As Emma notes, the bifolia are definitely out of order. That's why the scribes are mixed in the herbal as well as ff. 93-96 (the bifolium 93/96 is Scribe 1, and 94/95 is Scribe 3). The book was disbound quite early in its history (before the folio numbers were added), perhaps because of the catastrophic spill in the upper margin, and the bifolia were rebound in the wrong order. It is not possible to reconstruct the original order of bifolia and quires because there isn't enough evidence to make a convincing case. There aren't any incontrovertible offsets that pre-date the misbinding, for example, and there are at least a dozen leaves missing. We may have to wait until it can be read to reconstruct the original order. Or we may have to find a way to reconstruct the original order so that it can be read!

Currier might have been right about how the VM was put together, but he was wrong about the languages and about some assumptions he made about Eva DAIN and AM having the same meaning. 
I would not concern myself so much about different scribes, since copying one's works was the only way to duplicate them. Judged by the illustrations, the VM displays thematical unity that would hardly be possible if created by different authors, unless they were the monks in the same monastery. If the frequency of certain glyphs in the VM pages would be enough to assume a different author, then You are not allowed to view links. Register or Login to view. would be written by yet another author. It contains 24 glyph designated by Eva as M, which are relatively rare in the rest of the text. The explanation is simple: the highly inflective language  with many grammatical forms could look like a different language. I had examined many medieval writings and I didn't see one page where letter M would be as seldom used, which made me believe that Eva M was a ligature - a minim 'i' + Eva 'l'.  This is a Slavic verbal ending for a 3. person, singualar. For the letter M, I designated Eva IIN, which can also stand for other Latin letters - IW, IIV (JIV), WI. 
The Voynich words DY and EDY  also cannot support the theory of different languages, nor different authors: A writer or a poet can use different styles: for prayers and meditations, he can choose to write in a first person, or in a form of a dialogue - 1.st. per  and 2. person, present tense; if he wants to tell a story, he would use 3rd person. If he gives instructions, he would most likely use imperative mood, like in the recipes. 
In some languages, all these styles would require different endings. However, this does not mean that the same letters cannot stand for different grammatical styles, or at the end of the words that do not change with grammar.
To me, Currier was not helpful, but the Eva alphabet, although imperfect, was very helpful, because it pointed me where to search for the comparative manuscripts. 
You can find more detailed information on VM grammar on my blog