VMs: Re: Inks and retouching

[Jorge Stolfi <stolfi@xxxxxxxxxxxxx>]

Glen, 

  > [Glen:] Deductive reasoning at its best operates with a
  > conservative approach, carefully building one fact upon the other
  > until a picture emerges.
  
I don't think this is how it works. You certainly need data, and the 
more detailed and accurate the better; so your highly detailed 
transcription project is to be lauded.  However, you also need theories,
and the more of them the better.  Then you have to devise tests that
could confirm or disprove those theories. 

In fact, you need to have a theory even before you start collecting
data; otherwise you don't know whether to record the elephant's
weight, the length of its trunk, the number of flies on his back, the
length of its shadow, or the direction it is facing.  You need at least
a theory that tells which data is important and which is not.

You obviously spent a lot of time looking at stroke shapes and
apparent ink variations, and developed a detailed theory that explains
them as the work of a single author, modulated and modified by various
chemical and mechanical "natural accidents". According to your theory,
the shape of the plumes, and the colors of flowers, having been
determined by the author, are likely to be important; whereas the
precise shape of the vellum sheet is probably not important. Please
note that you *started* with a theory, and the way that you collected
your data was heavily influenced by it.

I haven't spent that much time on the topic, but until recently I
believed, like you, that the text was the work of a single person, and
that the variations in stroke density were accidental, as you say. I
had a different theory for the color paints, of course, and suspected
the Zodiac drawings of having been retouched; but I always thought
that the text was original. It was the color separation experiments
that made me consider retouching of the text, too, and the "daiin" of
f1r was one of the examples that convinced me.

So now we have at least two theories, just for the text and drawing
outlines:

  (1) only one author, who very rarely went back and retouched what he
  just wrote; stroke quality variations caused by mechanical and
  chemical accidents only.
  
  (2) many characters and drawing outlines, which had become hard to
  read or ambiguous, were retraced by later owner(s); many of the
  stroke quality variations are due to this retracing.

It will not be easy to decide which is correct, because both theories
have lots of adjustment knobs that can be twiddled to get around
contrary evidence. So we need to keep looking at more and more
evidence until either theory (or both) runs out of knobs.

I admit that theory (2) has an unfair advantage, since it accepts that
all inking and decay processes postulated by (1) are indeed possible.
So to disprove (2) one would have to show that *all* quirks of the
lettering are explainable as natural accidents, whereas to disprove
(1) it suffices to show *one* feature that cannot be explained that
way.

To make the game more fair, we can merge the two yes-no theories
into a quantititative one:

  (N) a certain number N of characters and drawing outlines,
  which had become hard to read or ambiguous, were retraced by later
  owner(s); this accounts for a commensurate fraction of the
  variations in stroke quality.

Obviously we both agree on theory (N); the only disputed point is the
value of N -- you seem to favor N<10; my current guess is N>1000
(dozens of characters per page were retraced, on a large fraction of
the pages).

You point out that many of my examples can be explained as natural
accidents. You may well be right on many of them. As I wrote in the
text below each image (which you are invited to read again), many of
the examples were presented as mere "suspects", meaning that, as far
as I can tell, they may *or may not* be retouchings. 

However, in many of the examples your explanations are not convincing
(see below), and I believe that there are enough of them -- in only
half a dozen pages that were examined -- to make a good case for a
large N.

I must point out that my survey was by no means exhaustive. I started
with f1r because I had already stared long at it. Then I picked *one*
page each of Herbal-A (f3v), Herbal-B (f26v), and Cosmo (f57v), at
random, and reported what I saw there. Then I checked f67r2 because of
the red text, and then scanned the Zodiac pages, where I knew there
were funny details on the drawings.

BTW, it was only when I got to the Zodiac that I noticed evidence of 
two separate retouching episodes --- which is not as strong as the 
evidence for *some* retouching, but seems quite pervasive on those 
pages. 

Our difference here is a good example of Bayesian theory: we see 
the same evidence, but since we have very different /a priori/ 
beliefs, we come out with different conclusions...

In an attempt to shake some of you /a priori/ probabilities, let me
note that even people who value the manuscript a lot may not share
your belief on the "sacrality" of the text --- and thus would not
consider retouching it a "sacrilege". We all know what Wilfrid Voynich
did to f1r, even though he believed that the manuscript was worth a
fortune. Now try to imagine Baresch, or some earlier owner like him,
who has been spending countless hours poring at the manuscript. Think
of how those faint strokes, which are hardly visible under Beinecke's
fluorescent floodlights, would look under brown candlelight. Why
wouldn't he have retraced those faint letters (very carefully of
course)? If father Petersen was willing to trace the entire text on
vellum paper, why couldn't Baresch have been just as thorough? So why
do you find the possibility of a Retoucher so unlikely /a priori/?

Now, with apologies to other list readers, let me reply to some
of your explanations:

  > [Glen:] The one [glyph on f1r] I pointed out was hardly visible,
  > yet showed absolutely no sign of "retouching". This qualifies as
  > "too faint to be read", so why wasn't it also "retouched"?
  
Perhaps because it wasn't that bad then?  

To me it is quite obvious that page f1r, in particular, suffered
considerable wear *before* and *after* the darker letters were
redrawn.
  
  > [Glen:] f1r-1
  > 4 - don't know if you're pointing to the 8 or the whole word, but no
  > "retouching" in this example is evident.

As explained in the the text below that image, the [4] refers to the
space between the two words, where my allucinating eyes see an
erased "s". (It is just an accidental observation, not related
to the Retouching issue.)

  > f1r-2
  > 1 - o8 - reinked pen - heaviest ink on the wider down strokes 
  > 2 - Hw - reinked pen - heaviest ink on the wider down strokes
  > 4 - o
  > 5 - 8 - heaviest ink on wide strokes
  
As said in the text, these look suspect, not certain. You may well be
right, or not. Time will tell, hopefully.

  > 3 - woe - reinked pen - all wide strokes
  
Not to my eyes, definitely not reinking. The ink is different, not
just thicker. But again, I have no solid proof (yet).

  > 6 - o
  > 7 - you say o8, but I see the 8 as darker than the o

Please read the text. I am not counting these, not even as suspects.

  > f1r-3
  > 1 - Wo - word initials - reinked pen
  
...with ink that, after writing one and a half glyphs, suddenly dried
up and changed color? 

  > 2 - o
  > 3 - o

Saying "it's an o" is not an explanation. The left side of the "o"[2] is 
exactly the same stroke as that of the "a" and "c" next to it. Why is
the "o" so much darker?  Even if the stroke on the right half somehow caused
more ink to flow, there is no sign of it having "flooded" the left stroke. 
Worse still for the "o"[3].

If the stroke on the right side of the "o" was the culprit, there
should be "o"s with light ink on the left, dark ink on the right; but
I didn't see any of those in my survey.

You seem to believe that the "o"s often look darker because the "o"
was done in a single stroke, by pulling the pen down on the left half
and pushing it up on the right half (or perhaps vice-versa?). Do you
have any evidence for this claim? I find it hard to believe; apart
from the difficulty of pushing the pen up without it catching on the
vellum, we should see more variation in width between the two sides.
On the contrary, on many "o"s, both halves are pressure-split
--- which as you know can hardly happen on a "pushing" stroke.

  > 4 - Wo89 - you think the word was retraced in its entirety

No, I beleive that the plume on the Wo is original, the rest was retraced.

  > I don't agree. The W is the end of a pen - faint. The o is
  > reinked, the 8 is a little lighter, and the 9 starts to fall off
  > in ink output. Transitional, not retouched.
  
I see those reinking effects too -- but all in distinctive dark ink,
i.e. by the Retoucher's pen.

  > 5 - continue 4 to 5, we see the h as probably the end of a pen,
  > and the compound glyph following the work of a reinked pen. The
  > plume was added after the ending 9 of the word is written,
  > somewhat like dotting an "i".

...and we close our eyes so as not to see the bits of light brown
ink sticking out from under the blotted "o".

  > f3v-1
  > 1 - you list as a [control]?
  
Yes.

  > Chemical burn on vellum, no pigment left. Age.

Since this page is very clean and shows no signs of wear, I find it
hard to believe that all the pigment fell off. But if it did, and the
"ink" we see is only a chemical residue or burn, that only makes the
contrast with the dark letters harder to explain. (Please don't tell
me that those still have the original pigment: why would it fall off
only from whole pen strokes, never from one half of a stroke only...)

  > 5 - down and to the right
  
...just like the other "i" stroke before it, and the bottom of the "8"
character after it. So why is this single stroke darker? (Guess:
because the original glyph, which can be seen under the dark ink, was
slightly bent. So the Retoucher, after puzzling over it, decided that
it was an "i" after all, not an "e" -- and "fixed" it so that he would
not have to puzzle it out again and again.)

  > 6 - Word initial
  
I found this one suspicious because the vertical stroke seems to end
abruptly and then a thinner and fainter stroke seems to extend from
under it. But OK, let's say that it was a skip of the pen...

  > 7 - down and to the right
  > 8 - down and to the right

In [7], a bit of old ink is showing on the left edge.

  > f26v-1
  > Please!!!!!

Please, please, read the text...

  > f57v-1
  > 1 - down and to the right
  
What?

  > 2 - o

Please read the text. There is lighter ink visible under the
dark one.

  > 4 - facial features - many things you didn't point out here.
  
Indeed, I was getting tired already...

  > Look at the long line, starting out dark, and fading toward the
  > end, especially when the pen changes angles. Transitional, not
  > retouched.
  
What I see is an entire figure, hair and all, drawn in faint lines.
(By your theory, the author would have done it all without a single
reinking of the pen, right?) Then there are three or four dark strokes
that could well have ben applied over faded ones: on the arm, under
chin, on the hand and ball. The latter were clearly meant to reinforce
an apparently important detail, in a part of the drawing that was
really faint (check the other hand near it.)

And how can you explain the "moustache"?
  
  > 5 - added on the end of an inking

As you wish... but *why*?

  > f67r2-2
  > You only have one place on this image that is possible retouching, #1.
  
Well, so there MAY have been a Retoucher after all.

  > Notice how the tail fades on a light pen when the angle changes.
  
No, notice how the original tail of the "9" continues, with a
different tone, after the retraced tail ends. Notice how the plume of
the "s", although thin and light, has the same blackish tone as the
body -- and looks quite different from the other plumes in this image.
Notice the two lines of text at left, all in light brown ink: by your
interpretation, those were written with a single inkload, correct?

All the best,

--stolfi
______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list