The Voynich Ninja

Hello there,

it has been already established that the Herbal section has Herbal-A and Herbal-B dialects present, which differ in several ways (You are not allowed to view links. Register or Login to view.).
One of the significant differences is presence of "shed"/"ched"/"eed"/"edy" and others, which I used to generate the pixel maps separated by quires
[attachment=13596]

The Quire 4 shows that folios 26 and 31 use Currier B,
[attachment=13597]
Quire 5 has folios 33,34,39 and 40 in Currier B,
Quire 6 has folios 41,43,46 and 48 in Currier B,
Quire 7 has folios 55 and 55 in Courier B,
lastly all 4 pages of Quire 8 are in Courier B.

So those quiers were most likely reorganized, and could potentially be two quires in Courier A , and two other in Courier B, which should simplify further statistical analysis.

So folks, this may be interesting, but most likely it is nothing...
I arranged those pages the following way. Quires 1, 2 and 3 stay the same (30 pages)
Then go Quire W(25,27,28,29,30,32), Quire X(35,36,37,38), Quire Y(49,42,44,45,47,56), Quire Z(51,52,53,54), those are just random placements that preserve the connected folios.
That gives 41801 words of a continuous Herbal-A text across 86 pages.
Then I built a Spearman matrix of the page-to-page similarity based on the most frequent bi-grams.

[attachment=13605]

So this one is interesting since it has the noticeable (and very regular) grid. The light gray noise pattern is the similarity between pages in range of 0.4-0.6 of the Spearman correlation coefficient.
The three dark lines for folios f14v, f27v, You are not allowed to view links. Register or Login to view. represent pages that are very different from the rest at 0.19.

Has anyone noticed anything like that before or is that one of those dull things that has already been reported a decade ago?