The Voynich Ninja

This is what looks a very interesting matter, although there's no clear picture at the moment.

I was thinking over Nick's post You are not allowed to view links. Register or Login to view. about tails of n having been written in a separate pass.

Indeed, in many places (not only 38v specifically) it looks like the tails of n were added in a separate pass. What betrays that in particular is that if we consider n as the combination of i and the tail, then those i-components which have a tail appended usually have a more or less prominent "hook" (to which the tail is then appended). The concept is illustrated by the figure (38v, line 7) below which shows the hook in blue and the tail in red.

[attachment=2271]

Note that the behaviour of non-vord-final i-shapes in vords ending with n is not unambiguous. They may have no hooks at all (38v line 5):

[attachment=2272]

They may have hooks less prominent than the vord-final hook (see fig. 1 above), which is, I guess, the most frequent case.

They may also have hooks as prominent as the vord-final hook. In this example of 38v, line 3, the hook of the next-to-last glyph is the same as that of the last glyph, the latter being not very prominent per se.

[attachment=2273]

Hooks of non-vord-final glyphs may be considered as natural connectors of several strokes of one glyph (e.g. like in Latin three "i" form an "m"). However, e.g. in this example (38v line 6), first two i shapes do have hooks (let's say, "connector hooks") and the fourth one also has a hook (let's say "terminal hook"), however the third one clearly does not have a hook:

[attachment=2274]

Of course, this just may designate that the sequence shall be read as "m" plus "n" (due to no "connector" between the two).

Let's put the issue of "connector hooks" aside for now and consider terminal hooks, and tails thereto appended. Some of the terminal hooks are very prominent, some are not. Clearly, they are not "connector hooks" in any case (because no i-shape ever follows). So there must be another explanation for the nature of terminal hooks. The simplest explanation would be that they are kind of embellishment, the result of the sweepy movement of the pen at a vord's ending. There are apparently no very many vords ending with i, if any at all, so this assumption is not very easy to check. The problem with hooks and tails within the n shapes, however, is that the vord is ending with the tail, it is not ending with the hook. So two questions manifest themselves:

1) Why two passes? Why not put the n shape in a single pass?
2) Why the need for the hook within n?

The simplest answer to the first question is that inscribing the n shape in a single pass is impossible from the writing technique perspective - the pen must be lifted from the surface in the midpoint. However, this is disproved by the existence of shapes such as b - which are clearly written in one pass.

The simplest answer to the second question is that the terminal hook is an "intermediary embellishment". The i with an "embellishing" hook is put down, and then the tail is just appended in the second pass. However, there are r glyphs also - which do have prominent hooks in much fewer cases. It is true that r sometimes has a prominent hook (see example below, 38v line 7), which sometimes makes it look even more like s, but it seems that on the whole, prominent terminal hooks in r are far less common than in n. I also have a suspicion that joining tail to the i not exactly at the top of i but slightly lower serves the purpose of distinguishing r (with their potential confusing hooks) from s.

[attachment=2275]

This direction of discourse leads one to suspicion that the terminal hook is placed as an indicator for to be used in the second pass - specifically, indicating that, by appending a tail, the shape is to be turned into n, not r. This is far not certain though, since, once again, r with hooks are there - albeit less frequent, they are quite numerous. So this is rather vague at the moment.

However, be it so or not, it is still not clear why bother with such "reminders" if the tail (be that for r or for n) can be appended in the first pass? You are placing a vord-final glyph. If you know that it should be r, place the r at once. If you know it should be n, place n at once. There is one evident explanation of this strange need for the second pass - that you do not know in advance whether it needs be r or n (or, say, l etc.). r , l and m clearly have the need to be written in two strokes, so they look just natural (in fact, many r's look like there are even three components within them, but that's another thread), but only n - that does not have that need - betrays these two strokes as two passes.

But there is something more which may either substitute this explanation of two passes - or even complement it. That's the "tail coverage", which is introduced by analogy with the You are not allowed to view links. Register or Login to view.. Indeed, one may notice that the tail of n (we'll come to other tails later) sometimes covers more than its own i, but also one or more preceding strokes (be they just i's or full EVA glyphs). Consider Fig. 4 above, where the tail covers four strokes, three of which do not belong to the n glyph. This occurrence is much like those in which the gallows coverage is manifested as something more that just pure embellishment. In here, there is little room between the lines for a sweepy curve with its natural radius to fit in - so the radius is made deliberately infinite, with the curve going horizontally - that is, in parallel with the baseline, in order to fit the "covering" tail in.

Sometimes the tip of the tail stops in between two preceding strokes so that it is not easy to determine the extent of the coverage (this is the case with some gallows also). There may have been some convention to resolve such ambiguities.

Can the tail of n cover anything beside just i shapes? Yes, it can - at least the glyph a . For that see e.g. You are not allowed to view links. Register or Login to view. line 7 or line 9 (at this point I'm tired preparing inline illustrations, sorry! ). Interestingly, sometimes the tail covers the whole a, while sometimes it covers only the i component of the a (see e.g. f42r, next-to-last line).

What about tails of other glyphs? Although seemingly not as often, the tail of r can also exhibit coverage - see e.g. f42r, line 2. So does the tail of s. The tail of s is very interesting in this respect. Exactly like r or n are commonly preceded by i or i-containing shapes, s is commonly preceded by e. Thus the tail of s likes to cover one or more e shapes - see e.g. f42r, last line. I have seen (don't remember where) the tail of s covering l - which means that there is no curve-to-curve (neither, presumably, line-to line, in terms of Cham's CLS) rule for the tail coverage. In other words, tails of curve-based glyphs can cover line-based glyphs and (presumably) vice versa.

The b, which also visually derives from e , can exhibit coverage as well - see e.g. f42v, line 10, where it covers two preceding instrances of e.

Returning to s, it sometimes demonstrates behaviour not very fitting into the described model of tail coverage. First, s is often seen as vord-beginning character, and in such cases it is not rare that it covers the space preceding the vord. I can't imagine any plausible need for that. Worse than that, s is sometimes seen as line-initial character, and in such cases it sometimes has a deliberately long tail of large radius - exactly as if it should have covered a string of characters - but there are no characters. Refer to You are not allowed to view links. Register or Login to view. line 10 for an illustration. Unless the long tail of s as vord-initial character has some special significance, this strange behaviour is something that jeopardises the whole concept of tail coverage. So, while I don't have a shade of doubt that gallows coverage exists, I'll leave room for doubt in the case of tail coverage.

What could be the significance of tail coverage? One thing that it would not be is marking covered i's as strokes of a single glyph. That's because coverage is observed over e's as well - and e shapes are clearly standalone, not combinable into a single glyph. Other than that, I can imagine some operator over the preceding glyphs as operands. What it is exactly - I can't tell, same as for the gallows coverage.

I have not researched the peculiarities outlined above systematically and limited myself (for the moment) to only several folios. Bottomline at the moment looks to me as follows:

a) terminal hooks - not clear if they have any special significance. May serve as markers for converting i to n, but that's not very likely due to abundance of examples to the contrary. You are not allowed to view links. Register or Login to view. line 8 is a vivid example of how i with a prominent tail hook was converted to m, not to n. Most probably they are sort of embellishment (or, rather, outcome of a particular writing technique) for the vord-ending i-strokes before the tail of any sort is appended. What is important is that the presence of terminal hooks in n shapes betrays the two-pass nature of creating the flow of script.

b) connector hooks - not sure if they are really meant to combine i-strokes into a single glyph or not. Really an open question. More consideration needed.

c) tail coverage - more chances that it does have special significance versus simple embellishment. If so, it does not serve to mark stroke combinations as single glyphs. Might be some operator.

d) two-pass process - reflection upon why it is needed in the first place might be the clue to understanding Voynichese. A two-pass process is a paradigm fundamentally different from the simple (or not-so-simple) substitution (which is usually the heavy focus of Voynich-deciphering attempts). The two-pass paradigm also largely negates the idea of decomposition of Voynich glyphs as "base-shape+tail" constructs (as discussed in You are not allowed to view links. Register or Login to view.)

The very division of the process in two passes means either one of the two following things, or both at the same time:

a) the scribe does not know in advance which exactly tail is to be appended
b) the scribe does not know in advance the extent of coverage to be introduced

Either of these means that the scribe does not know in advance the final look of the vord he is putting down.

Wow, this has been a long post. Thx for your patience.

Anton, I have a HUGE chart documenting the different versions of "daiin" that includes the different tail coverages. I don't know if posting it is a good idea though. It is HUGE and it's a work in progress.

I suppose maybe I could post the first dozen folios, although even that is huge (it's about 47" wide by about 24" long PDF just to represent the first folios).


To give you an idea of what I mean by huge, here is a screensnap of a tiny little corner of the chart:



Down the left side, it lists folios. Along the top it lists the different versions of daiin. For example:



As examples (there are many more, this is just to give the idea). They are actual clips from the manuscript and I looked very very carefully at the length of the tails and how far over the minims they stretched. I also included the other glyphs that precede "ain" because there are several that typically precede it, not just EVA-d and they tend to be overlooked.

The dain group is MUCH more varied than is recorded in the popular transcripts, which means that many of the statistical studies might be worthless. IF the length of the tail is meaningful (in addition to the number of minims), then important information is being overlooked.

Also, a separate issue is whether the minims are single strokes or, in some cases, double strokes.


I think some of them might be double-strokes (like a "u" shape). The reason I think so is that early in the manuscript, the scribe expressly put a cap over two minims and distinguished them from the following "w" shape with a tail. It isn't proof, but it does increase the probability that some of them are double minims (and some seem to be drawn that way), which is why I tried to record single and double minims separately as ii or u.


Edit [addition]: I found an example. I don't want to assume too much, but it's almost like the scribe is indicating some relationship between the first two minims and the last two minims:

[attachment=2280]

I'm not the expert here, but if, all of a sudden, the quill is moved in a different direction, the ink flow may be different, so a darker stroke, or a lighter one that needs to be corrected afterwards, may not be so strange.

Quote:Anton: What could be the significance of tail coverage? One thing that it would not be is marking covered i's as strokes of a single glyph. That's because coverage is observed over e's as well - and e shapes are clearly standalone, not combinable into a single glyph. Other than that, I can imagine some operator over the preceding glyphs as operands. What it is exactly - I can't tell, same as for the gallows coverage.

There are many ways of looking at this and one is that there is some special VMS system at work here (just as there appears to be some consistency to the way curves and lines are combined).


So, I offer this information as how it TRADITIONALLY works, even though I know it might not be how Voynichese works. So, it may or may not be relevant to Voynichese, but this is what occurs in most medieval Latin script...


In languages that use Latin scribal conventions, the direction of the tail, and the length of the tail, and the presence or absence of a hook (by this I mean an end hook, not the connecting hook) is meaningful for some letters and not meaningful for others.

• For example, in Latin, EVA-r and EVA-s can have tails of varying lengths and it doesn't change the meaning.... usually. There are occasional exceptions. Most of the time EVA-s stands for "con" (if it's c with a tail rather than e with a tail, both of which are drawn the same way) but it can sometimes stand for "contra". Most scribes don't change the length of the tail, but some of the more methodical ones will give EVA-s a longer tail if it's "contra".
  
• Some scribes add a hook to the end of the tail if the letter that has been left out is "r" rather than "m" or "n". This end-hook IS found on some of EVA-r characters in the VMS. It may not have the same meaning in Voynichese but it definitely has the same specific shape and it sometimes hints at this shape at the ends of some of the dain groups.
  
• For EVA-y the length of the tail doesn't change the meaning, only the position in the word changes the meaning. The same is true for EVA-m, the direction of the tail doesn't change anything but unlike EVA-y which is commonly used at both the beginnings and ends of words, EVA-m is usually at the ends of words (and sometimes is used as a line-end or paragraph-end marker).
  
• For some scribes, the length of end-tail (as when it is attached to the letters "a" or "u" where it is often seen, similar to the tail on the vord "dain") is meaningful. An end-tail is a connected macron, connected to make it easier to write quickly. Most scribes don't worry about the length of the tail but a few will end the tail above the place where the missing letters would be located (this is often used to indicate a missing "r"). Some will use a short tail to mean one or two letters are missing and a longer tail to mean several letters are missing.
  

In other words, in Latin, the length of the tail and the end-position of the tail (and the shape at the end of the tail, if there is a hook) are meaningful, but it is only the methodical scribes who do it this way. From what I have seen, about 80% of scribes don't bother with these distinctions or some will make distinctions for some letters and not for others.

I have examples of these (and also clips of where they occur in the VMS), but it's very time-consuming to hunt them up and post them.

Tails as a separate pass...

Here is the abbreviated word principatus:

[attachment=2277]

Notice the "smoke" macron (here it is simplified) that usually stands for er/re/ir/ri is used to mean "ri", and the tail that represents "-us" has been added separately (not in one stroke).

.
In the VMS, sometimes the "ain" group has the tail added without lifting the pen. Multiple scribes have worked on it, but I have not had time to study whether one connected the tails more than others, but here are two examples, in two different sets of handwriting. Each adds the tail as a single stroke:

[attachment=2278]  [attachment=2279]

I'm not sure that your example to the right is in single stroke, but there definitely are examples where the single stroke is certain. E.g. here (76r line 9):

[attachment=2283]

the n is single-stroke (the s is double-stroke).

On the other hand, there are examples when two strokes are literally disjoint, e.g. 66v line 11:

[attachment=2284]

Maybe, as Rene says, this is the result of a particular writing technique. In this case, two strokes would not mean "two passes", but just a convenient way of performing the only pass. The fact that b-shapes do not  exhibit two-stroke behaviour would then be explained by that they do not have the salient point (the point where the quill "changes direction").

About the tail coverage, it definitely waits statistical approaches, although I'm not sure what's the appropriate way to approach it. Simple counts won't tell us much - whatever the distribution, it always can be written off to the writing habits. Perhaps, it makes sense to pick some more or less frequent vord with a tail n (luckily, daiin, as the most frequent vord, presents itself as a perfect candidate), an then see if vord combinations featuring this vord do exhibit any correlation with the tail coverage of n. Say, if vord X is frequently found before daiin with no tail coverage, while vord Y is frequently found before daiin with double coverage, that would look interesting, suggesting that the two daiins stand for two different words.

If you look at the image below, which shows a selection of the text from f38r, it's clear that there's a pen problem.

[attachment=2285]

On every line there is a glyph with a stroke which transitions suddenly from light to dark ink. Note kain in the last line where the loop of k goes dark when it reaches vertically upward and then light again when it reaches vertically downward. The n in the same word also goes dark when it reaches the same point vertically upward.

There are actually three kinds of strokes, where tails are concerned...

• single stroke (no pause). This is the most difficult. Coordination has to be good to do this and still get the shape right (if it's done too quickly, the shape gets muddy or overly rounded).
  
• single stroke (slight pause). This is pretty common. You pull the quill in one direction and there's a slight pause as you change the direction of the quill. There might be a miniscule elbow at the point where the pen direction changes, but it's still basically one stroke because the pen doesn't get lifted.
  
• double stroke. This is also quite common. The pen stops completely and there is a slight or complete lift of the pen before the new stroke is begun.
  

(30-07-2018, 06:33 PM)Emma May Smith Wrote: You are not allowed to view links. Register or Login to view.If you look at the image below, which shows a selection of the text from f38r, it's clear that there's a pen problem.



On every line there is a glyph with a stroke which transitions suddenly from light to dark ink. Note kain in the last line where the loop of k goes dark when it reaches vertically upward and then light again when it reaches vertically downward. The n in the same word also goes dark when it reaches the same point vertically upward.

Good example, Emma.


Where the three tails go up and to the left, it is darker, and there is "pen stutter" on the bottom two. This scribe was probably right-handed and probably was slowing down when drawing strokes that go in the "contrary" direction (contrary to what is comfortable for that person's hand and for the angle he or she was holding the quill). If the scribe is not comfortable going in the contrary direction, it also results in blobs, parts that get filled in.

There are a lot of indications that the VMS was not written by high-level scribes. This is one of them.