An unfortunate circumstance came to my mind. The fact is that many folios are not flattened in scanning, so in fact that would influence the observed baseline and may give false impression of the change in its trends. Probably that would not affect the jumps, but there will be false positives as to changes in slope.
Consider e.g. the upper part of f7r. You are not allowed to view links.
Register or
Login to view.
This is a thought provoking thread, Anton! What appears to be going on is that the collection of words that the scribe needed to write onto the folio were split into groups. The groups were then written down in lines or part lines at (x,y) coordinates on the folio. Presumably there were rules s/he was following: these words have to go top left, these other words have to be at the bottom right, these have to be between the branches of the plant, and so forth.
As you have shown, on some folios, the lines of words appear disjoint, and this is where a different group was being written, just like in your image with the blue and black underlines.
My question is: how might this relate to the folio-positional frequencies of the glyphs. Some glyphs "prefer" being at the top of the folio, others at the left of the folio, etc.. Does this somehow tie in with the folio placement of groups of words you have shown? Are we able to isolate the disjoint word blocks and look at their glyph frequency properties, for example?
Hi Julian,
I have not undertaken any systematic analysis, that's a lot of work indeed, but from what I see at a glance, the major pattern observed in many places is as follows. There is the "main" block, which is the middle one. It is generally written top to bottom and is the largest in size. Next, there is the left block comprised of one or two vords each line. It's not easy to tell whether it's written before the main block or just "filled in" afterwards. And next, there is the right block which is written BTT, and thus it postdates the main block. In some cases it looks like the right block is written not in one pass in itself. Beside that, the main block does not in all cases look contigious, that is it looks like there are blanks left when the major part of it is written, which are filled afterwards.
About plotting the glyph density. That's an interesting idea, but with two problems.
First of all, it's a question how to approach that. Different folios have different amount of text, and, beside that, it's not clear what is to be considered a "container" across which the plot has to be made. By "container" I mean a well-defined range of folio space which is filled with text in accordance with pre-defined rules, and which will set forth the coordinate space for the plot. For example, should the plot be made across the entire folio, or paragraph-wise? It's not clear if the entire folio is a container, or each paragraph is a container in itself. There are cases which look as if the last line of the preceding paragraph was written after the first line of the subsequent paragraph, due to the baseline jump in the last line to avoid the gallows intrusion. However, there are also cases when the basejump (if I may, for brevity) of the last line is exhibited without any intrusion from beneath, which may suggest that it's just an effect of multipass in general, but not of the gallows intrusion in particular.
Next, even if each single paragraph is a container, what about paragraphs split by imagery, e.g. by a stem of a plant? Are they one container, or two different containers (e.g. two different paragraphs, effectively)? I'd say, in certain cases it looks so. If the latter, then is each of them written in mutipass in itself?
The second problem is that I'm not sure offhand what may be achieved by such analysis, except highlighting some things already known (e.g. that the m likes to be line-ending). One could hope to detect the systematic block borders through such analysis, but if the block borders are defined in spatial terms (like, e.g. each tenth character of each third line is the block border) then this analysis will show nothing, because with the natural flow of text the position of the tenth character may be occupied by any glyph. If, on the other hand, the block borders are defined by certain markers, then, by definition, such markers will be observed in different positions (if you have the positions already defined, you don't need markers), and the density plot will again show nothing. It will reveal something only if the borders are certain markers in certain systematic positions - which is, as I explained, redundant, but not impossible for the "naiive" medieval mindset.
Regarding the coordinate system, when I looked for the structure of narration back in 2015, I introduced L-coordinates and P-coordinates (short for "line" and "paragraph", respectively), wherein the value of the coordinate is the ratio of the number of the line (or paragraph) in question to the total number of lines (or paragraphs) in a folio. One could do the same for characters in a line. I'm not sure offhand, if it makes sense to use relative (as I did) or rather absolute coordinates in the analysis that you propose. Maybe it makes sense to use both.
Hi Anton,
Yes, agreed that the positional analysis is cumbersome. One could define (0,0) as the position of the top left glyph on the folio. Each glyph on the folio would have a bounding box, in millimetres or pixels, whatever is more convenient. And each glyph would be offset by (x,y) from the top left glyph. Similarly, each word would have a bounding box, comprised of the envelope (probably convex hull) of its glyphs' bounding boxes. And, the various blocks would also have bounding boxes, or polygons. With this data one could ask how densely packed words are in a given block, and one could make stats on which glyphs appear in which folio locations or word blocks. I feel like I've seen a similar proposal (to include absolute positional info in a transcription) somewhere years ago, perhaps by Rene?
Ideally an image analysis library could do all this work - a folio by folio positional analysis of glyphs, words, blocks - but it seems computationally challenging. Perhaps one of the adversarial neural nets would do a good job? Or, find a willing student to do it by hand, LOL.
Julian
OK, I see what you mean - coordinates linked to actual geometry, not to units of text.
That would show the text density and potentially help to assess which pieces text were laid down before or after.
But that would reveal little in terms of underlying rules, because the amount of source plain text is, most probably, different for different folios and different paragraphs.
Columns were usually respected and kept simple in medieval manuscripts. There are a few examples of rambling columns (e.g., some of the northern Carthusian manuscripts).
There are also some less common examples with staircased columns (they go partway down the folio then move over to the right and continue). Since it would take months or years of hunting to find one of these if you were explicitly searching for it, I thought I would post an example that I have in my files, for the record:
[attachment=4413]
I know, it's not the same as multipass, but it doesn't hurt to know that less conventional columns did exist in the Middle Ages.
Very interesting example!
I've been considering whether the Voynichese text may render a two (or several) column original into one-column pattern, but bottom up is something really excessive to that.
Interesting..
However..
I find it more compelling to think that, writers being sloppy and pages being unruled, this is a case of fitting the gallows to the above word, which happened to be slightly high by sloppy accident.
I accidentally write some words slightly higher often enough, even on ruled paper!
As for word spacing certain places being strange.. Well. If it's a copy from a draft or other text, and clearly the drawings came first, one might expect the end of the line being slightly more cramped.
I'm reminded of a biblical manuscript where the scribe, realizing his mistake, towards the last third wrote progressively denser so as to fit a preset manuscript length.
(28-07-2020, 09:21 PM)ddskbnbn Wrote: You are not allowed to view links. Register or Login to view.I find it more compelling to think that, writers being sloppy and pages being unruled, this is a case of fitting the gallows to the above word, which happened to be slightly high by sloppy accident.
How about baseline jumps within vords themselves?
(04-05-2020, 06:50 PM)Anton Wrote: You are not allowed to view links. Register or Login to view.In fact, many vords starting with o followed by gallows have their baseline jump up starting from the said gallows.