Different parchment color through Herbal section?

[quimqu]

I was worried about the same thing Jorge said, that each photo could have its own exposure. So I tried to check it with the Beinecke images themselves.

Every Beinecke photo shows not only the page but also a thin strip of the facing page, on the other side of the fold. 

   

The photo of You are not allowed to view links. Register or Login to view. shows the gutter edge of f25v, and the photo of You are not allowed to view links. Register or Login to view. shows the edge of f26r. So inside one single photo you have two different pages with exactly the same exposure, wich is what we need.

Two things you can do with that:

• compare the page with the strip of its neighbour inside the same photo, so the camera can’t play any role;
  
• compare the same strip of parchment in the two photos where it appears, wich gives how much brighter one photo is than the other.
  

There are 55 openings in the herbal where I can measure this, and 19 of them have an A page on one side and a B page on the other. Results in L*, B minus A:

measure	B − A
photo against photo (like Koen)	+0.86 ± 0.40
inside one single photo, page against facing strip	+0.87 ± 0.50
exposure of the B photo against the A photo	−0.06 ± 0.30

So the B photos are not more exposed than the A ones, and inside one photo the B parchment is still lighter by the same amount. The exposure diference between the two photos of an opening is also no bigger than the measuring noise. And there is no sign of automatic processing: I measured the black background and is is at the same level in all 183 photos, nothing is burnt to white, and the parchment doesn’t get brighter on pages with more ink and paint. Unluckily Beinecke removed the metadata of the images, where we could see exposure and more interesting data.

With only 19 mixed openings, the inside-one-photo number alone is not significant, but it is the same size as Koen’s. What this can’t rule out is a slow drift along the book, but in quires 4 to 7 A and B alternate from one opening to the next, so a drift can’t follow the language. Together with npcompl33t’s check on the multispectral images, I think the lighter B parchment is real and not beeing made by the photography.