The Voynich Ninja

I see no evidence of retracing anywhere in this image (no double strokes), but I do have a question. If I understand your interpretation correctly, the retracer added the following strokes to t (in red on the right), is this correct? Why not retrace the legs in full, if both their upper and their lower part needed fixing?

[attachment=12886]

(10-12-2025, 10:05 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.I see no evidence of retracing anywhere in this image (no double strokes)

What about the leg of the p?

Quote:Why not retrace the legs in full, if both their upper and their lower part needed fixing?

Retracingtext or drawing is much slower and tiring than tracing it the first time.

The original Scribe could "let his hands do the writing".  The Retracer had to look constantly at the original trace and move slowly in order to follow exactly the same path.  So it makes sense that the Retracer would redo only those parts of the glyphs that really needed it.

You can see that in @nablator's example from f4r, line 3: why did the Retracer (who you claim is the Scribe himself) retrace only the C of that Ch, in the middle of a line?  Why not retrace the whole Ch?

(The only talk about user interface design that I watched and was actually useful explained that the feedback loop from eye to muscles to hand takes (IIRC) about 100-200 milliseconds, even though it takes only 10-20 milliseconds for the brain to process what the eyes see.  From that fact one can deduce, among other things, that the time it takes for someone to click a button that pops up on the screen is proportional to the log of distance from cursor to button divided by the size of the button -- and one can estimate the constant factor too.  

One simple experiment to confirm that 100 ms time is to draw two parallel lines and then try to draw a zigzag line between them, as fast and wide as one can, without crossing the lines.  Drawing the zigzag itself does not depend on the eyes, so it can be very fast; but adjusting the width of the zigzag depends on feedback through the eye, so the envelope of the zigzag ends up being a wavy line whose period is those 100 ms.)

All the best, --stolfi

(11-12-2025, 05:00 AM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.What about the leg of the p?

There is a clear double line or a loop there and there is some difference in the density of the ink on visible scans, but not huge. Given that this is the first letter of the page, I'd list possible explanations in order of falling plausibility to me:

1) This was an attempt to embellish the first letter of the page. The glyph is huge and a thin leg may have looked not good enough. There are other examples of embellished first gallows of the page in the MS, including clear wide features made of double strokes.

2) The first vertical stroke turned out a bit faint due to half empty/half dry quill or unmixed ink, similar to other known retracements at the beginning of a line, so the scribe decided to redo it.

3) This was a work of a proper retracer who decided that for some reason the leg needed fixing, but not the top horizontal bar, for example.

Do you really think option 3 is the most likely?

(11-12-2025, 05:00 AM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.

Quote:Why not retrace the legs in full, if both their upper and their lower part needed fixing?

Retracingtext or drawing is much slower and tiring than tracing it the first time.

Actually I think it's not. If we are talking about normal retracing that leaves obvious double strokes and that is attested in a few places in the Voynich MS and in other contemporary manuscripts, it's an easy job, you don't have to think about having enough space, what word to write, etc, so this is not much slower or even faster than writing. Basically this is a "follow the strokes" exercise that most kids perfect in elementary school or earlier.

If we are talking about mystical perfect retracing that leaves no unequivocal evidence, yes, this is a hard job in your eyes, and in my eyes it's practically impossible in the scope of the whole manuscript, so should be easily rejected by default.

(11-12-2025, 05:00 AM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.You can see that in @nablator's example from f4r, line 3: why did the Retracer (who you claim is the Scribe himself) retrace only the C of that Ch, in the middle of a line?  Why not retrace the whole Ch?

Note that I think this example proves my point. The scribe there redid full complete strokes, even putting another horizontal bar on top of the existing bar. I was asking why the retracer in your example retouched each leg by using two short separate strokes instead of using normal full strokes.

(11-12-2025, 01:13 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.Do you really think option 3 is the most likely?

 
Not less likely thna your 1 or 2...

Quote:Actually I think it's not. If we are talking about normal retracing that leaves obvious double strokes and that is attested in a few places in the Voynich MS and in other contemporary manuscripts, it's an easy job, you don't have to think about having enough space, what word to write, etc, so this is not much slower or even faster than writing.

That could apply to the Scribe himself, because he would not have cared whether the retracing was visible.  It could be the case of the daiin on f1r.  But the hypothetical Rt1 restorer was hired to restore the manuscript as accurately as possible, so he would have tried harder to follow and cover the original traces.  And more so on the text than on the figures.

And even for the Scribe himself, retracing would definitely have been slower than writing, even if he tolerated some parallel tracing.  The shape of his glyphs varies a lot from word to word.  Again, when writing he could have "let his hand do the writing" at full speed, but when retracing he had to use slower traces and eye feedback.

Quote:If we are talking about mystical perfect retracing that leaves no unequivocal evidence, yes, this is a hard job in your eyes

But, first, there are many cases of darker glyphs in the middle of lighter text.  You would consider them "backtracing", cases of the original Scribe himself going back and retracing glyphs that he had just written; and I can agree that many of them are just that.  But they show that it is quite possible to retrace many glyphs without leaving "unequivocal evidence" other than the darker ink...

And, second, there are in fact thousands of cases where the Rt1 Retracer left evidence of his work. Like bits of Rt0 traces poking out under his traces, plumes retraced in the wrong direction and therefore with wrong width, glyphs turned into weirdos, double traces in illustrations that make no sense, ...  

Quote:Note that I think [@nablator's] example proves my point. The scribe there redid full complete strokes, even putting another horizontal bar on top of the existing bar.

But the ligature is normally part of the h, not of the C...

By the way, was the upper right loop of p retraced like this? Why was the retracer so obsessed with making these t and x shapes at the intersections of two strokes in many places on this page? In my simple basic all original ink world these short stumps of dark ink are easily explained by either some of the dark ink mixing with still fresh faint ink of the first stroke or the quill holding the weak ink picking and carrying a bit of the fresh dark ink along the way of a weak ink stroke.

But the retracer focusing on the intersections of the strokes, where there should be a double layer of the original ink anyway, doesn't seem a bit strange?

[attachment=12894]

(11-12-2025, 01:45 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.But, first, there are many cases of darker glyphs in the middle of lighter text.  You would consider them "backtracing", cases of the original Scribe himself going back and retracing glyphs that he had just written; and I can agree that many of them are just that.  But they show that it is quite possible to retrace many glyphs without leaving "unequivocal evidence" other than the darker ink...

1) Retouching fresh wet ink is totally different, because wet inks will mix by themselves. In some cases you don't even need to follow any stroke at all - you can just touch any part of the existing stroke with a tip of the quill and the new ink will fill the existing stroke with new shade. Or if you do make a new stroke, the inks from the new and the old stroke will mix leaving no clear traces of double strokes, just a single somewhat wider stroke.

2) In many cases I usually don't consider dark patches of ink as a sign of retracing. As I said before, I consider most variation of the ink density to be the actual variation of the density of the ink, something like oily/gooey blobs of dark ink suspended in relatively transparent ink.

About the example with c and h that @nablator has provided, the dark strokes were not added immediately, you can clearly see a sharp border between the dark and the faint horizontal bar, the inks didn't mix there. This is what proper retracing looks like. However, the left e part is quite wide and could have plausibly covered the original stroke entirely. This is not impossible at all, I'm not arguing that there should be a double stroke each time anything is retraced. I'm arguing that there should be plenty of these obvious double strokes. Like in this example: for two strokes one perfectly hides the original*, the other almost doesn't hide the original at all.

*unless the original was just a h, but I don't think I'm ready for this discussion now, and it's not really important here.

(11-12-2025, 01:45 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.And, second, there are in fact thousands of cases where the Rt1 Retracer left evidence of his work. Like bits of Rt0 traces poking out under his traces,

If you mean like clearly obviously poking, I'd like to see them. If you mean like faint parallel lines perfectly mimicking the shape of the main stroke, these could be:

1) image compression/processing artifacts
2) small part of the original ink that managed to seep into the vellum and reacted with it under the surface
3) lens/matrix artifacts
4) the side of the nib of the quill touching some bumps on the vellum
5) for poorly mixed ink with dark blobs - just the center of the stroke executed with a blob that sticks to the middle of the quill and the side filled with weak free flowing ink.

For a stroke to be obviously a different trace poking from under a later trace it should obviously differ in shape and not be perfectly parallel/offset copy of the new stroke.

(11-12-2025, 01:45 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view. plumes retraced in the wrong direction and therefore with wrong width, glyphs turned into weirdos, double traces in illustrations that make no sense, ... 

I think by now you know my answer to this one, but I don't mind repeating: I don't consider appeals to "wrong" glyphs, shapes or traces valid until we have some means of knowing the intention of the original work. It's clear that this manuscript is unusual, no matter retraced or not, putting an arbitrary line where its original weirdness should stop and some extraneous weirdness starts does not work like a good argument for me.

(11-12-2025, 01:45 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.

Quote:Note that I think [@nablator's] example proves my point. The scribe there redid full complete strokes, even putting another horizontal bar on top of the existing bar.

But the ligature is normally part of the h, not of the C...

This doesn't seem to matter here, I have no strong opinion about the origin of this particular feature, I don't mind it being a work of a later retracer or the original scribe. My point is that whoever did this obvious overwriting, used full natural complete strokes and not straight dash lines that your interpretation suggests in the gallows above. And yes, this clearly looks like something written over the original text some time after the original text was done, with at least enough time passing for the original ink to properly dry out.

(11-12-2025, 01:48 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.By the way, was the upper right loop of p retraced like this?

In general, the theory is that Rt1 retraced parts of the text and drawing that were "too" faint, but left some parts that were still OK at the time.

The original scribe traced gallows and puffs (excluing the "ornate" ones) in two strokes: first the left leg, from top to bottom, then the "head" and the right leg.  The legs were traced with the pressure used in other "broadstrokes", like the e and i strokes of Ch, Sh, l, iin, etc.  The horizontal parts were traced with reduced pressure, to reduce the change of the quill snagging into the vellum. The the loop(s) were traced in the counterclockwise direction (like the plumes of r, s, Sh, n), so that the top right (NE) quarter of the right loop was traced in the worst direction, SE to NW; and hence had to be traced with even less pressure.  

As a result, the NE part of the right loop was often the faintest part of the gallows. We can see that in those cases where the whole gallows appears to be original, with no retracing.

Therefore, it is perfectly understandable that many gallows that just that part of the right loop retraced by Rt1.

Quote:Why was the retracer so obsessed with making these t and x shapes at the intersections of two strokes in many places on this page?

There is a "hydraulic" explanation, see below. But another explanation is simply that those bits of the loops were indeed "too" faint, and so the Retracer fixed them.

Also beware that, in these crude pseudocolor images, any ink that is light red, yellow, or orange looks darker when it runs over the "blue grain" texture.   I hope to correct for that with more sophisticated multispectral analysis...

Quote:In my simple basic all original ink world these short stumps of dark ink are easily explained by either some of the dark ink mixing with still fresh faint ink of the first stroke or the quill holding the weak ink picking and carrying a bit of the fresh dark ink along the way of a weak ink stroke.

This thing does seem to happen, but not just with fresh ink.  Even dry ink, containing gum arabic, should be more wettable than blank vellum. According to the Net, the final stage of vellum making included  rubbing it lightly with powdered You are not allowed to view links. Register or Login to view., a hydrophobic resin.  I suppose the point was to prevent the ink from spreading out (as seen on You are not allowed to view links. Register or Login to view. and another page I forgot, where presumably this resin was not applied).  Thus if a sufficiently charged quill was drawn across some previous ink stroke, no matter how old, I expect that some of the new ink would bleed sideways into the old one, for a short distance.  

On the VMS, this effect seems to happen also when paint was applied over inked lines.

Quote:But the retracer focusing on the intersections of the strokes, where there should be a double layer of the original ink anyway, doesn't seem a bit strange?

The double layer would be at the intersection of the two old traces, and that was of course covered when one of them was retraced.  The details you are considering are bits of old trace that are adjacent to the intersection. Those would not have a double layer.

All the best, --stolfi

More hallucinations in psychedelic color on page f8r.  Mostly the usual stuff.
[attachment=12901]
This clip spans the middle part (around the plant's stem) of lines 9-12 of page f8r. As usual it shows what seem to be at least three layers of ink: original (Rt0), a gneral restoration (Rt1), and at least one set of scattered retracings (Rt2).
   
    This is a false-color image created by combining three narrow-band images from the Lazarus 2010 set -- MB940IR_012, MB700IR_008, and MB535GN_004 -- as the red, green, and blue channels of an RGB image.  No processing was performed other than affinely scaling the samples in each channel to the full 0-1 range, with ImageMagick's "-normalize" directive.  The fuzzy blue areas are bleedthrough of the green paint from f8v.

    (A) Original (Rt0) parts of the p. (B) Original traces of the plant outline.  (C ) The tip of the plume is Rt0; the rest is Rt1, retraced closckwise.
   
    (J) Restored (Rt1) parts of the p. (K) Rt1 traces of the plant outline. (LQ) The left leg of the t and the tail of the l are Rt0 or Rt1; the rest of the word is mostly Rt2.
   
    (P) These glyphs are presumably Rt2.

One more from page f8r:
[attachment=12902]
This clip spans the left end of lines 9-12 of page f8r. It is a composite of the Lazarus narrow-band images MB940IR_012, MB700IR_008, and MB365UV_007 as the red, green, and blue channels.  No processing was done other than scaling each channel to the full 0-1 range.
   
    As usual this clip shows some surviving original (Rt0) ink strokes, several strokes from the general restoration (Rt1), and a few later retracings (Rt2).
   
    Besides the "blue grain" texture, another confounding factor in the interpretation of this clip is that the vellum has horizontal warps spaced ~10 mm apart.  Because of the oblique lighting used, all colors become lighter or darker depending on the slant of the surface at each point.  This effect can be seen in the areas of blank vellum.
   
    (A) Most of this t is Rt0, with small touches of Rt1. (B) The left e stroke of this Ch shows a bit of Rt0 extending from under the Rt1 retrace. (C ) The two e strokes of this Sh are Rt1, the rest is Rt0. (D) The tip of the tail is Rt0, the rest is mostly Rt1. (E) A bit of the Rt0 stroke is visible at the top of each of the i in the aiin. (F) This a had an Rt0 foot connecting to the first i stroke, which was omitted by Rt1. (H) Heads of both ps have at least 5 strokes of Rt1 each.
   
    (J) Glyphs completey retraced by Rt1. (K) Surviving bits of Rt0 show that this glyph was originally a plain o, but Rt1 made it ambiguous, almost an a.
   
    (Q) On this t the lower half of the left leg is Rt0,  the loop and top of the right leg are Rt2 (probably in 3 strokes), and the rest is Rt1. (R,S,T) Presumably Rt2-retraced glyphs.

This is the bottom line daiin from your last image with attached mm rulers from MSI scans. Do I understand it correctly, that you suggest that the dark parts of the ascender of n are added separately by a retracer as at least 3 different strokes?

[attachment=12903]