The hidden structure of the VMS

[quimqu]

@npcompl33t, nice tool. Your symmetric pairs (f1v/f8r, f36v/f37r…) are all on the inner face of the bifolio, the two pages you see together when a single sheet lies open. That fits what I found in the text: the inner pair shares more vocabulary than the outer pair, and the outer pages use more t than k*. I think the sheets were written folded, like a small booklet.

You have the data to test it on the layout:

• Use 4-page bifolia of the same hand and section, with text masks measured from the vellum edges (not aligned to the first line).
  
• Compute the overlap for each pair type: inner (a-v|b-r, also mirrored across the gutter), outer (a-r|b-v), same leaf and diagonals.
  
• As a null, shuffle the page positions within each sheet.
  

If the inner pair beats the others, that’s a physical confirmation of how the sheets were written.

One caveat: paint from the drawings often bleeds through the vellum, so the two sides of the same leaf (a-r|a-v) may match mirrored for physical reasons. That should be kept in mind when reading that pair, and the masks are better built from the text lines than from the image areas.

*The inner/outer results aren’t published yet. I’ll post them soon.