Posted 2004-07-15, updated 2004-07-22.

Evidence of text retouching on f1r

It has long ben suspected that the Voynich Manuscript was modified in some way after it was first written.

That conjecture was often made about acessory items: the colored paints that were applied to the drawings, the page numbers, the bits of European script (such as the "michiton" text and the Zodiac section month names), and the various "key sequences".

In the new high-reslution color scans made available by the Beinecke Library, one can now see clear evidence that the main text of the VMS was itself touched-up or corrected at a later time. This is an example, from page f1r:


On the first word of the second line, "daiin" in EVA, it can be seen that the final "iin" was retraced in dark ink over a fainter original. Such evidence is visible also on the following word "shcthey where the "sh" and the "y" were visibly retraced on top of lighter strokes.

Does this happen elsewhere?

Judging by the color of the ink, the bottom half of the initial "d" may also have been retraced, as well as the "c" at far right, and "ol" and "od" on the next line -- although the presumed original strokes are not visible there.

The existence of two distinctive tones of ink have been noted by other people on other pages, e.g. by Gabriel Landini and several others who have used his pigment-separation technique. In the case of the Zodiac month names, for example, the difference in tone was noticeable ven on the black-and-white images. In fact, hints of such retouching (if not conclusive evidence) is seen all over the manuscript, as shown below.

As for f1r in particular, evidence of such retouching is visible all over the page. Even the big red weirdos seem to have been first outlined with the light brown ink, then redrawn with the darker brown ink, and then filled-in in red.

Natural variation?

It has long been observed that, on f1r as well as on many other pages, some characters are noticeably darker than others. These differences were commonly attributed to natural variations in stroke density as the pen periodically runs out of ink and is refilled.

In hindsight, though, that explanantion seems unlikely. For one thing, the period between refillings seems too variable, and often too short -- sometimes a couple of characters. Although I have never handled a quill pen, my experience with improvised reed pens makes me believe that a properly made quill pen should hold enough ink for a couple of words at least.

Furthermore, if that were the explanantion, one would expect to see the strokes getting gradually lighter from left to right, then abruptly darker; but one often sees instead an almost binary situation, with isolated dark characters in the middle of fairly uniform light text. See for example the "ol" of "chol", on the third line of the above image.

While the "natural" explanation is not completely ruled out, in my view, the alternative explanation -- that those characters were overwritten at a later time -- seems at least as likely.

When was the touch-up done?

Apparently the retoucher's goal was to restore characters that had become too faint to be read. This would indicate a long period between the original writing and the retouching. Moreover, to my eyes the touch-up ink is just as dark as that of the page numbers -- whose style, according to some analysts, suggests that they are a century or more later than the VMS itself.

On the other hand, the retouching appears to have been done with great care: only rarely does one see the underlying strokes. Moreover, most touched-up symbols definitely look like the original ones, so it would seem that the retoucher was familiar with the Voynichese alphabet. (To understand this argument, try to copy a sample of a script you do not undarstand, and show it to someone who does.)

Thus the retoucher may have been the author himself. But it may also have been someone who had been studying the manuscript for some time (Baresch, perhaps), and to whom (like us) the script was not just a jumble of pen strokes, but rather a string of recognizable symbols.

Another observation that may be relevant is that the touch-up ink and the original ink have exactly the same hue and saturation -- a fact pointed out by Gabriel and confirmed by other people, including myself. The only difference between them is their darkness ("value"). Moreover, both inks seem to be somewhat peculiar -- more like a yellowish-brown opaque tempera paint, than the purplish black "iron gallate" ink that was commonly used to write on vellum. Although these features may be an artifact of image processing, it is also possible that the ink is indeed a suspension of an insoluble pigment resembling ochre or "yellow earth" (a very common pigment), and that the dark ink is either based on the "burnt" (roasted) version of the pigment, or was mixed with another black pigment (such as soot).

What are the implications?

The evidence of generalized touch-up on the text, if confirmed, is somewhat bad news for text analysts. For one thing, if the Retoucher was not the author himself, we do not know how accurate he was; he may have mistaken EVA "a" for "o", "ih" for "ch", and so on.

There is also the possibility that, by the time he decided to restore the text, many parts had already faded out beyond recovery. Indeed, the light text is often barely visible.

Can we recover the original text?

Gabriel's ink separation technique can be used to analyze a typical VMS text image into a mixture of three "pigments": "vellum" (the background color), "ochre", and "soot". The following example comes from page f4r:


original
 

"vellum"
RGB = 233 228 213

"ocher"
RGB = 172 140 098

"soot"
RGB = 116 097 074

Assuming that the "new" ink differs from the old one only in darkness, the "ochre" image above shows both the "old" and "new" ink together, while the "soot" image shows the black pigment present only in the "new" ink, together with the shadows due to the irregular vellum surface.

Note that the pigment separation algebra can easily distinguish an intrinsically darker pigment from the apparent darkening of strokes due solely to changes in ink thickness, e.g. after refilling the pen (the "natural" explanation discussed above): the latter shows up only on the "brown" image, while the former shows up only on the "soot" image.

On the other hand, one cannot separate the black pigment from the shadows by color algebra alone, since these two factors have exactly the same effect on the image colors. However, as the "soot" image shows, we can tell these two apart easily by their texture -- the shadows are fine-grained random noise, while the black pigment is organized as strokes.

How pervasive was the retouching?

Hints of retouching, similar to those seen on page f1r is visible throughout the manuscript. Here are some examples, from a few arbitrarily chosen pages.

Page f1r


f1r-1

Note the "o" in "choldy"[1], and also the curved part of the first "c" in "cthar"[2].

These are only "suspects" of retracing, not "convicts", because the presumed "old ink" layer is not visible. However note the contrast betwen them and the adjacent letters, and compare the color of "o"[1] with that of the "o"[3].

Note that the ink seems to be just as thick in both "o"s -- that is, except for a few bald spots, the pigment seems to cover the vellum surface completely in both cases. Yet "o"[3] is quite a bit lighter. The difference *could* be due to the brown pigment being partly transparent, or to nonlinear scanner response -- but it fits rather nicely a two-ink model.

The "c" in "cthar"[2], besides being darker than surroundin strokes, is also misplaced relative to the horizontal bar, and not as curved as other "c"s in light brown. Note that the rest of the "cth" has been quite damaged, so it is possible that the original version of "c"[2] was barely visible then the retouching was done.

A few days ago, while I was trying to separate colors for the "big red weirdo" page, I tried hard to separate the "brown ink" from the "shadow" component in this part of the image, but the "o"[1] would always show up strong on the "shadow" layer. Until then, I had assumed (like most Voynichologists) that the light/dark variations were mostly due to variable ink thickness. It was this problematic color difference between the "o"[1] and the nearby characters that got me looking for evidence of retracing.

By the way, the mottled apeparance of the all letters on this page (both old and new) seems to be due to the "orange skin" texture of the vellum. My interpretation is that the bare spots are raised parts, where the ink has been rubbed off over time. These bald spots, mostly lacking on internal pages, may indicate that f1r was for a long time the book's cover page, even after the Retoucher did his job. Perhaps this clue can help us decide whether the sections were once separate booklets...

To add to the mystery: I fancy that I see the ghost of a small "s"[4] on line 3, in the space between the initial "s" and the next "d". Perhaps it is a Rorschachism, but on the next image there is a bona-fide "s" of just the same size...


f1r-2

My chief suspects in this image are the "o d"[1], "cthh"[2], "chol"[3], "o"[4], and "d"[5]. The "old ink" layer may be just barely visible in "chol"[3].

Also suspicious are the "o"[6] and "o d"[7], but the the green bleedthrough makes it hard to build a case against them.


f1r-3

The "sho"[1] is quite interesting because of the strong contrast between the base and the plume, and the backward "c" stroke on the base. It could be a very rare weirdo... or it could be that, at this point, the Retoucher was not yet familiar enough with the alphabet to correctly guess the shape of that stroke.

The "o"[2] of "choaiin" has a slip-of-the pen at the top which is rarely seen elsewhere.

The "shody"[4] was probably retraced in its entirety, except for the plume. There seems to be a bit of the old ink visible under the leftmost "c" stroke. Considering that the "ch" just below it was apparently spared by the Retoucher, the old ink must have been very faint around here -- so it is not surprising that we do not see more of it.

The "o"[3] in "oky" is suspicious for its color, placement, and shape, but there is no smoking gun.

The next word[5] is one case where the Retoucher apparently messed things up. My guess is that the original word was "cthshhy" or "ckhshey" or something in between. However this whole area seems to have suffered a lot of abrasion, so presumably the first "c" was completely lost (do I see a faint ghost of it, closer to the preceding "y"?) and the rest of the platform gallows was damaged. The Retoucher omitted the first "c", retraced the gallows as "k", and then replaced the "h" by a ligated round blob ("-o" in froggished EVA). Note that bits of the original "h" are still visible. He also connected the "-o" further on to the first "c" stroke of the "sh", which he retraced too. The result is a transcriber's nightmare...

By the way, the replacement of "c" strokes by "-o" or "o-" is visible in a few other pages that I have checked. Again, I would thnk that at this point the Retoucher was still not used to the rules of Voynichese "orthography".

The two "o"[6,7] on the bottom line are worrisome because they are almost halfway between "o" and "a". Perhaps the Retoucher could not figure out the original letters, so he made the new ones ambiguous on purpose?

Page f3v


f3v-1

This is a Herbal-A page. Unlike f1r, it looks very clean and shows no sign of abrasion, presumably because it has always been buried inside a quire.

The "old ink" pigment in this image looks different from that on f1r. Not only it seems to have a reddish tinge, but also it seems to be a transparent liquid, rather than an opaque powder. Even in the thicker strokes, such as the "t"[1] on line 5, it seems to only modify the vellum color, rather than obscure it.

These differences should not be taken too seriously because they may be artifacts of image processing, such as contrast/brightness changes and gamma correction. (That is why technical images should always include a color reference chart!) For one thing, the "new ink" and the vellum itself seem to be "red shifted" with respect to f1r...

The fading of the old ink does not seem to be due to wear (as on f1r) but to the pigment losing its color. That, together with the transparency, suggests an organic dye rather than a mineral pigment.

My chief suspects for retouching are

There seem to be "old ink" visible in second "i"[5] of daiidy and the first "c" of "shol"[7].

Several other dark charaters on this image are suspects too, but let them pass for now.

Page f26v


f26v-1

This is an Herbal-B page. The page seems clean, presumably for being inside a quire. The ink color again seems different -- sepia brown, very watery. There are many darker characters but they are also obviously over-inked; so it is hard to tell whether they are retraced or just inking accidents.

Nevertheless, I suspect retracing because in a few cases the "old ink" seems to be visible. Check the "okeo"[1] on line 3 (which may have been "okey" originally). Also the "cpchedy"[2] on line 4 may have been "cphhedy"; there seems to be an "old ink" ligature partly hidden by the spreading of the "p" ink.

Note the strange word[3] on line 1 ("ai-'he" in froggished EVA). Could it be a Retoucher mix-up? If so, what was the original word?

Page f57v


f57v-1

This is the page with the "417 key sequence". There are plenty of retracing suspects on this page (although no really obvious smoking gun). To mention just a few:

The plume on the "r"[1] seems to have been clumslily retouched. It is worth noting that on this page one sees another weirdo charater (not visible here) that looks like an "r" but with a half-length plume. So the plume on this "r"[1] may have been enhanced in order to distinguish it from that weirdo.

By the way, two instances of that weirdo character, elsewhere on the page, seem to have been redrawn with an intensely black ink -- which is the expected appearance of good iron-gall ink (as seen, e.g., in Kircher's correspondence).

Bits of the old ink may be visible under the the bottom "o"[2], although it is possible that the pen just went dry on the NE part of that character.

There are four instances of the fancy "p"[3] on this page (and only on this page, it seems). In all four instances, the flourish is suspected of retracing. In the other three, the flourish was written with a single stroke. In this instance, however, the final loop seems to be a separate stroke. Perhaps the Retoucher drew this one first, and mistakenly assumed that the final loop was a separate glyph?

The dark "moustache"[4] on the figure is quite peculiar, given that the hair is very light (reputedly "blond", but perhaps just "faded brown"). My guess is that the Retoucher did not notice the original mouth[5] (thought it was a dimple on the chin?) and so he added a new one. (Such "over-restoration" of drawings occurs on several other pages.)

Page f67r2


f67r2-1

The big question on this page is whether the red text is original or not. Unfortunately the red strokes are visibly thicker than the brown ones, and would have completely masked them. There may be a bit of brown visible below the ligature of the second "sho"[1], but don't quote me on that.

This page is quite clean and we have a good sample of the red paint (the blotted "k"[2]), so perhaps color separation may be able to reveal a brown component in the parts where the red is not fully opaque.

However, even if there is brown text uderneath the red, it may not count as a Retoucher's job. If the author wanted this text to be in red, for emphasis or decoration, he may have written it in brown first, as sort of a sketch, then retraced it in red.


f67r2-2

Apart from the red text, there are only a couple of retouching suspects on this page -- including the The final "sody"[1] above.

The "s" plume in that word seems to be in dark ink too. So its rather strange shape may be the result of retracing a nearly invisible original amidst the natural spots of the vellum.

Speaking of spots on the vellum, I just noticed what seems to be an erased sentence, starting with "or" above point [2] and ending with "n" above point [3]. Another Rorschachism?

Note the "s"[4] and "y"[5] on "soeey" at left. I can understand why a jealous owner like Baresch would want to restore the faint letters before they faded out of sight.

Page f71r


f71r-1

This is the second Aries page; together with its sibling, they seem to have the most primitive-looking drawings in the book.

At first I though that this page showed very few signs of retouching; my main suspect was the decorated square at 10:30 on the second text band (not shown), and the non-voynichese "aberil" month name (not shown).

However I then noticed a very, very faint "ch"[1] between two words. So now I suspect that, in fact, the entire page has been retraced -- text and drawings! However this retracing either used an ink that was lighter than the "new ink" seen on other pages (and in the "aberil"name), or else the "new ink" has, for some reason, faded a lot more on this page than elsewhere.

This suspicion applies to the whole of the Zodiac section (which, by the way, seems to be the oldest one in the book). There may have been two Retouchers -- which may of course be the same person, or even the author himself -- acting at different epochs.

It seems that only a few of the framing circles have been retraced. Moreover, while the circles were originally traced around some template, they were retraced free-hand. So, the obviously unretouched circles like[3,4] may be telling us what the original ink looks like now: "all but invisible"...

This "ancient retracing" may be responsible for the very weird weirdos seen on the label[5] ("opaldar"? "opals ar"?).

Page f72v3

On this page, the (Second) Retoucher apparently added only the month name and "corrected" a few drawings. Here is a detail that I always found rather supicious (in a different sense 8-), which only now I managed to understand.


f72v3-1

In the original drawing, the nymph[1] at the top had very thick legs, close together. The Retoucher apparently though that she was meant to have thinner legs, spread further apart - so he "restored" (rather clumsily) the "missing" left edge of her right leg.

The Retoucher also added or retraced the outline[2,3] of the right breast on both nymphs. In fact, it seems that many nymph drawings in the Zodiac section were originally missing that detail (and a few of them still do), but were "fixed" by the Retoucher.

Page f72v1

As in other Zodiac pages, most of the text and drawings here may have been retraced by the First Retoucher. (The outer framing circle[6] may show what remained of the original ink.) The Second Retoucher may have added the month's name, and made scattered only "fixes" to the drawings. Here are some of the worst examples of his "art":


f72v1-1

On both nymphs[1,2], the eyes, breasts, and pubic area were "improved" rather childishly. The hairline on the temple was reinforced, and headgear was apparently added -- some sort of bonnet[3] for the right nymph, a crown[4] for the left one. He also smeared some "new ink" over the text[5].

I see no sign of "old ink" under the crown, but perhaps there were very faint traces, and the Retoucher just guessed that it was a crown.

On a previous zodiac page, the dark ink was limited to the tiaras or crownlets worn by several nymphs. Since no "old ink" was readily visible, I though that perhaps the dark paint had been intentionally applied by the author, and may have been originally some other color, or perhaps the sticky glue that was used to to hold gold leaf in place. However the "corrections" above to the nymphs anatomy, which were apparently made with the same ink, largely dispell that explanation: those tiaras, too, mau have been a contribution of the Second Retoucher.

Page f73r


f73r-1

The lower nymph[1] is one of the few cases where the Second Retoucher failed to add a rigth breast (which he did on the nymph above[2]).

The Second Retoucher also retraced half of the labels in the outermost star band, including the two[3,4] shown above.


f73r-2

In this image we may be seeing the three layers of ink -- by the Author, in very light brown, as in the framing circles[0], the "okary"[1] label and the outlines of the middle nymph[2]; by the First Retoucher, in medium brown, as in the fourth framing circle[10] (hand traced over a fainter original?), the text bands[3], and the central star[4]; and then the Second Retoucher, in dark brown as in the "okeos"[5] label and various details on the nymphs[6,7,8,9].

There are many bits of old ink visible around the strokes the "okeos"[5] label.

Last edited on 2004-07-22 06:12:06 by stolfi