A disquieting thought

[obelus]

Another example of single-scribe variation is the prominent excess (new to me) of EVA:al in f58, which is not observed in q20 (Starred), although both are attributed by LFG to Scribe 3.  On the left is a density map of EVA:al in paragraph-style text over the entire ms, with rows indexed by line number and columns indexed by first-character position in the line:

   

The stripe of high density at lines ~1400-1500 is f58, while q20 spans lines ~3200-4200.

MarcoP's example of EVA:eol is mapped on the right.  Big Pharma covers lines ~2900-3000, while the entire Herbal section spans lines ~0-1600.

The resolution is 50 lines x 2 characters, or 100 EVA characters/bin.

[pfeaster]

Encoding does not have to be deterministic.

One case in point: lately I've been studying Pitman phonography (i.e., phonetic shorthand) and reading a number of older texts written in it, both by people who were very adept at using it, such as Isaac Pitman himself, and by people who were not.  Phonography wasn't intended to be difficult to read -- quite the opposite -- but it offered multiple strategies for handling particular situations, and different writers favored different ones, while the habits of individual writers sometimes varied over time or by context (e.g., personal notes versus correspondence).  Historically, new forms were added over time to make it possible to write common phoneme combinations more efficiently; for example, a single distinctively-shaped loop was introduced after a few years to represent /str/, which would previously have taken more strokes to write.  Sometimes an innovation was introduced and then withdrawn (e.g., two rarely-contrastive vowels were assigned different marks only briefly in the mid-1840s).  The formal differences among texts in Pitman phonography don't seem any less extreme than the formal differences among Voynich languages, scribes, quires, etc.  But Pitman phonography is both meaningful and legible, and all the examples I have in mind are in English.

[Anton]

The encoding mechanism may be the same while just some parameter differs.