Over the past few years, I've developed a system for evaluating text more objectively based on mathematical values assigned to letters so they can be searched and sorted by the software rather than through human guesswork. It takes into account letter forms, pen thickness, degree of connectedness, letter spacing, line spacing, slant, proportion of descenders/ascenders, and a few other factors.
As I mentioned, the project ended up being much bigger than I originally anticipated because I wanted to do it right rather than simply looking through the scripts and making a subjective evaluation.
I felt it was worth doing because when I first looked at the VMS, it looked to me like the differences in hands was more than simply someone writing at a different time with a different pen (which is significant because it suggests a group project and common knowledge of the writing system) and I wanted to confirm or deny this impression through careful research.
(14-02-2017, 08:20 PM)Koen Gh. Wrote: You are not allowed to view links. Register or Login to view.While writing my latest blog post, I read on Rene's site You are not allowed to view links. Register or Login to view. that Currier concluded that six to eight scribes wrote in the manuscript. That's main text, ignoring marginalia. I have no experience with telling apart different handwritings, so I thought I'd ask here.
Has this been confirmed? Has anyone else studied this in detail?
Of how many hands can we be certain? Again, only considering core text, not marginalia or quire numbers.
And how do the different hands correspond to Currier A and B?
For Currier A word types similar to [daiin] and [chol] are typical whereas for Currier B types similar to [chol] are rarely used. Instead many types which are rare or doesn't exist in Currier A are used in Currier B. Typical word types for Currier B are words like [chedy]/[shedy] or words starting with [qo] like [qokain]/[qokaiin] or [qokeedy]/[qokedy].
Lets compare the statistics for words similar to [chol] and [chedy] in Currier A and Currier B.
In Currier A the words [chol], [shol], [sho] and [chy] are frequently used. In Currier B everything changes. The words [chol], [shol], [sho] and [chy] are less frequent then in Currier A. Instead the words [chey], [shey], [cheey] and [sheey] are frequently used.
Currier A - statistics for words similar to [chol] and [chedy]:
char ( 30) chear ( 22) cheear ( 1)
chal ( 15) cheal ( 7) cheeal ( 1)
chor (182) cheor ( 59) cheeor ( 8)
chol (280) cheol ( 71) cheeol ( 3)
cho ( 48) cheo ( 12) cheeo (---)
chy (104) chey ( 78) cheey ( 34)
chdy ( 9) chedy ( 2) cheedy ( 3)
chody ( 44) cheody ( 27) cheeody ( 4)
Currier A - statistics for words similar to [shol] and [shedy]:
shar ( 12) shear ( 5) sheear ( 2)
shal ( 1) sheal ( 2) sheeal (---)
shor ( 69) sheor ( 26) sheeor ( 4)
shol (118) sheol ( 37) sheeol ( 5)
sho (106) sheo ( 15) sheeo ( 1)
shy ( 65) shey ( 60) sheey ( 39)
shdy ( 2) shedy ( 2) sheedy ( 1)
shody ( 26) sheody ( 15) sheeody ( 2)
Currier B - statistics for words similar to [chol] and [chedy]:
char ( 36) chear ( 23) cheear ( 1)
chal ( 27) cheal ( 22) cheeal ( 1)
chor ( 27) cheor ( 31) cheeor ( 4)
chol ( 89) cheol ( 88) cheeol ( 6)
cho ( 12) cheo ( 33) cheeo ( 16)
chy ( 31) chey (238) cheey (122)
chdy (119) chedy (470) cheedy ( 52)
chody ( 42) cheody ( 46) cheeody ( 8)
Currier B - statistics for words similar to [shol] and [shedy]:
shar ( 19) shear ( 15) sheear (---)
shal ( 13) sheal ( 12) sheeal ( 1)
shor ( 18) sheor ( 24) sheeor ( 1)
shol ( 55) sheol ( 64) sheeol ( 9)
sho ( 14) sheo ( 20) sheeo ( 7)
shy ( 32) shey (193) sheey ( 92)
shdy ( 38) shedy (401) sheedy ( 78)
shody ( 24) sheody ( 25) sheeody (---)
Quote:And are we even certain about the clean distinction between these "languages" to begin with?
There is no clean distinction between Currier A and Currier B. The following table shows the frequencies for some words typical for Currier A like [daiin] and [chol] and for Currier B like [chedy], [qokaiin] and [qokeedy]. By doing so it is possible to order the sections of the VMS in a way that two sections with similar frequencies follow each other. This way it is possible to demonstrate a steady development from Currier A to Currier B.
section daiin aiin qokaiin chol cheody chedy shedy qokeedy total word count
Herbal in Currier A 403 33 1 228 8 1 0 0 8087
Pharmaceutical (A) 99 39 2 45 18 1 1 0 2529
Astronomical 23 38 0 8 8 4 0 0 2136
Cosmological 36 56 18 19 7 24 17 4 2691
Herbal in Currier B 72 72 20 13 7 62 35 9 3233
Stars (B) 122 193 114 62 33 190 113 137 10673
Biological (B) 84 32 88 14 0 210 247 153 6911
There is no difference between Currier A and B regarding the usage of spelling variants. In both "languages" words common on one page are maybe rare on the next page. The statistics can even differ for the two sides of the same folio. The only pattern which can be found on every page of the VMS is that they contain similar words like [daiin] and [aiin] or [chol] and [chor]. Typical for Currier A and B is that an additional gallow glyph was added in front of the first word for a paragraph. There is also the same preference to use line initial glyphs like [y], [o], [d] or [s]. The method for creating the text is the same for the whole manuscript [see You are not allowed to view links.
Register or
Login to view.]. Therefore I would say that there was only one scribe.
Quote:Torsten
...Therefore I would say that there was only one scribe.
I would interpret that differently. I'm pretty confident there's more than one hand in this manuscript and I'm pretty close to being able to demonstrate that.
I would prefer to say there was only one method (even if it has variants) rather than only one scribe.
Yes, even if you think the MS is some complex cipher or procedurally generated, it is still possible for others to have learned the method, like JKP says.
Same goes if you think it's a copy of a premade example, or a transcription of a foreign script. Any theory allows for multiple scribes to use the method consistently.
Unless of course one wishes to argue something like flow of consciousness or glossolalia. In that case multiple scribes would disprove the theory.
(16-02-2017, 12:22 AM)-JKP- Wrote: You are not allowed to view links. Register or Login to view.Quote:Torsten
...Therefore I would say that there was only one scribe.
I would interpret that differently. I'm pretty confident there's more than one hand in this manuscript and I'm pretty close to being able to demonstrate that.
I would prefer to say there was only one method (even if it has variants) rather than only one scribe.
Indeed. Currier clearly addresses two different observations:
- the presence of different statistics in the text (languages A vs. B)
- the presence of different handwriting properties (hands 1 vs. 2, but also 3,4, X, Y...)
On the languages, he clearly states his criteria, and already in the earlier days Jim Gillogly made the effort of double-checking his identification based on these criteria.
The presence of more or less complete transcriptions makes it possible to automate the identification of languages or dialects according to all sorts of different criteria.
With the handwriting it is quite different. We don't know his criteria. There is also no numerical data (at least publicly available) to automate such an identification. I will be very interested to see the results of JKP, because this seems to be one of the things he has been working on.
One of Currier's main points was, in addition to the above, that he saw a full correlation between the application of different hands and different languages. This is why people may be inclined to connect the two.
However, I am not sure that this correlation has been demonstrated sufficiently.
It is clearly obvious (visually) in the alternation of Herbal-A vs. Herbal-B pages, but elsewhere it is not so much the case.
I think that it is very important to keep this uncertainty in mind.
(16-02-2017, 08:08 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view....
One of Currier's main points was, in addition to the above, that he saw a full correlation between the application of different hands and different languages. This is why people may be inclined to connect the two.
However, I am not sure that this correlation has been demonstrated sufficiently.
It is clearly obvious (visually) in the alternation of Herbal-A vs. Herbal-B pages, but elsewhere it is not so much the case.
...
This is one of the things that particularly interests me (the possible correlation between hands and Voynichese dialects. I'd like to have some hard numbers associated with the letter forms and am developing a method that can be mathematically defended, so that any connection between the two can be studied more objectively.
I've made no subjective attempts so far to correlated hands and Currier languages and have not recorded any of the Currier designations (or even looked at them). I plan to use the numbers to get a sense of what's going on.