The Voynich Ninja

Hi all, 

This is not a proposed solution. 

I had a slow day at the office and wondered if I could get the Curve-Line System fully working, and I could! .. admittedly it has the "something something too many freedoms" issue and I am working on refining it, but I thought it would be interesting to share something that works anyway. 

One possible benefit this system would have (if it can be shown to be accurate) would be defining spaces that are not obvious, as in if the next glyph would break the system it is more likely to be a space than not, the same could be said for non-obvious glyphs either due to them being done a bit scruffy, or damage to the original glyph in one way or another.


How the system works

1. The default state of the vellum is C (not \)

2. \ can't be added without a switch or transition

3. A switch is only activated by a non-switch glyph (aa oo ao oa etc are all treated as a single switch, not 2)


(Glyphs have EVA names)

Switch. 
Represented by ">", Function: switches text from C to \ or \ to C
Glyphs - A O



Forward transition. 
Represented by "\C", Function: \ followed by C without requiring a switch
Glyphs - FKPT(CH)(SH)DRS
(Gallows + Bench + D R S)

Benched gallows are treated as \\CC


Reverse transition. 
Represented by "C\", Function: C followed by \ without requiring a switch 
Glyph - Y

All other glyphs are treated as C or \ depending on if they are constructed on top of a C shape or a \ shape. 


Word examples

Daiin = \C>\\\

D = \C

a = >

i = \

i = \

n = \


Shol = \C>\

Sh = \C

o  = >

l  = \
 

Qokchy = C>\C\CC\

Q  = C

o  = >

K  = \C

ch = \C

y  = C\



Otshol = >\C\C>\

O  = >

t  = \C

sh = \C

o  = >

l  = \


Line example - I used 26r as I have the MSI image (Line 1). 


\C\CC>\CC\  >\C>\\\C  C>  C\  >\C\C>\C  \CC\\C\CCC\  C\\C\CC\CC\  >\\  \C>\C>\C\\CC\CC\

So it works, but as you can see it works because most of the text is made up of transitions (bold) which are more fluid.. freedoms. I will now keep transition status for the gallows glyphs and EVA: D, but set all others to C or \. 

CH - CC
SH - CC
R - \
S - C
Y - C

Benched gallows are treated as C\CC

\CCCC>\CC  >\C>\\\  C>  C  >\CCC>\C  CCC\CCCCC  C\CCCC\CC >\\  CC>\C>CCC\CC\CC   

The red text is where the system breaks. It could just be that this is 2 words and not 1, it is a very long word for Voynich text "chofochcphdy" and a unique word (only 1 match using Voynichese website), the break would make this "chofo" "chcphdy". "chofo" has no matches but there is 1 match for "chofol" and the page next to 26r (26v) has "chof" as a unique word, which, well who knows if that means anything but it is there. "chcphdy" has 2 matches, You are not allowed to view links. Register or Login to view. and f85r1, so maybe there is an argument that this is 2 words and not one, which I suppose shows rather nicely what I am working towards.



There are many more examples of this system not working I spotted while coming up with the framework for something that would work, I have a lot more work to do on this but I am putting it down for tonight and I just wanted to share what I have. I will update the thread as I go.

If you have any thoughts, please let me know

Having and answer and constructing evidence to support it / Having evidence and constructing an answer to support it.


In my first post I wasn't unaware that the whole thing was pretty unscientific, but I just wanted to see if this theory could work more broadly by amending it.
Now I want to dismantle it a bit and see if there is any sort of supporting "evidence" I rather to just think of it as logical steps which can be followed.

Lets looks at EVA: "r" and "s".

If you tilt this normal r backwards so the stem is at around 45', stick a curve on the bottom of it and you have s, sort of, right? Anyway, this is basically
how to tell apart "r" and "s" in Voynichese, "s" is curved at the bottom, "r" is not. Which sounds easy enough but.. hand writing.

From the below I think its fairly simple to see "r" is made up of a "\" line and "c" (backwards) curve.



In the system I propose "r" is made of "\" followed by "c".

This should mean that "s" is made up of "c" followed by "c", following the same logic.



Example of both on a page



Looking at You are not allowed to view links. Register or Login to view. (as it has very clear writing, and I happen to be on it already), do "r" and "s" work in this way within the system?

r - Conformance to system - 13/13 - r has an "o" or "a" switch when following from a "c shape" glyph and is followed by a "c shape" or switch then "\ shape"
s - Conformance to system (includes EVA: sh) - 20/25 - excluding sh - 9/13

3 of the cases I wouldn't be too sure of as they are next to plant drawings and may have unobvious spaces, 2 seem clear.
All 5 cases are caused by "ls"

So does s not work this way, or does l not work as "\"?

Maybe it reflects what the l glyph looks like also, a straight line-curve combo

If this were the case "ls" could not be preceded by a glyph ending with a "c" shape - c, e, s, h, q, f, p, k, t, g also all sh, ch, sh, cth.. etc, but that is taken care of by "h".

Search results of entire manuscript 
Total results for "ls" - 445. (of which "ls" as a stand alone word - 10)

cls - 0
els - 0
sls - 1
hls - 4
qls - 0
fls - 0
pls - 0
kls - 0
tls - 0
gls - 0

Factchecking
sls - incorrect transcription, it is "rl s"
hls - all 4 cases caused by "chls". 3 cases are unclear due to spacing, might be "chl s" (chl as stand alone word - 26. s standalone 244.) You are not allowed to view links. Register or Login to view. contains a clear "chls", this case would not work with the system.

So I think that shows that "l" is more likely to function as "\" than "c", but it can't only be "\" if it is often followed by a glyph starting with a "c" shape - c, e, s, h, q, g

lc - 742
le - 62
ls - 445
lh - 0
lq - 2
lg - 15 (+4 more noted below)

If l is "\C" as the above seems to suggest, we should not see results for l followed by a glyph which is a "\ shape" - m, n, i. However transitional "\ shape"
glyphs should be fairly common in comparison - f, p, k, t, d - I will include r and l also, assuming they are from the above.

lm - 8
ln - 0
li - 1

Factchecking
lm - lg,lg,lg,im,lg - 1 case not found. 2 cases are "lm", they are both line-end in recipes
li - la

Transitional glyphs
lf - 39
lp - 40
lk - 1079
lt - 107
ld - 452

lr - 40
ll - 28


Ideally I'd check the all the conforming instances too.. but.. time. 

Conclusion, "l" works as starting with \ and ending with c. 
It is likely "r" also works this way and "s" starts with c and ends with c.. but I will do similar tests using these glyphs before coming to a conclusion on that. 

I'm sure it will all come crashing down at some point, but for this little bit of work it was nice to see the glyph shapes just followed the function idea.

I just wanted to give a small update before heading off to family for xmas

Glyphs are currently as follows

Switch - ">"
a, o

Line - "\"
i, n, m

Curve - "c"
c, e, g, h, q, b

Double curve - "cc"
ch, s

Triple curve - "ccc"
sh

Line-curve - "k"
t, k, p, f, l, r, d

Transition - "c>kc" (benched gallows)
c-gallows-h

"Wildcard" - x
y



You are not allowed to view links. Register or Login to view.
kxk>\\x xkcc>k k>\\\ >kcc>k xkcc>\\\ c>kc>kccx
k>kcc>kx cck>k cc ccc>k c>kcx c>k>
c>k>\\\ c>kccx x cc>k ccx xkx cc>\\\
cc>\\kx cck>k kx c>kcx k>kcc cc>k>\\\ k
>k>cc>k (otochor) >k ccc>k>\\\ cc>k k>\ xkcc>\\\ k>\
cc>\\\ k >\\ k k k>k cc>k xk>kkx (ytoldy) kcc>k kccx c>kcx
ccc>k c>kcx k>\\\x cc>k k>\

kxk>\\ ccc>\\\ c>x cc ccc>k k>k>\ xkccc >kcccccx k>\\kkx (daiildy)
kkccccc> (dlssho) k>k cccccx c>kcx xk>kx cc> cc>k xkx k>\\\ k>\\k>k
c>kx cc>k>\\\ ccc>k ccc>k ccckx (sheky) k>\\\ c>kccx kc>k cc>\\\ cc>\\\
xcc>\\ k>k ccx k>k>k ccc>\ k>\ >kcc >\\\ cccccx c>kc>k
>k>k ccx cc>k c>kc>k x>k >\ cc>\ cc>\\\ ccckx (chety) ccxkx cc>k
ccc> xkccx cccx k>\\\ ccc>kcx

Non-conforming words noted above in brackets.

otochor - is likely "oto chor"
ytoldy - words with "ld/dl" are an issue currently
daiildy - might be "daiil dy" but again "ld/dl" problem
dlssho - very likely "dls sho", "ld/dl" issue again
sheky - "ek" and e+gallows in general is another issue
chety - same as above e+gallows issue


There are many more issues with the system beyond this page awaiting Currier B's "ed" for example.
My current thought is that e+gallows might be a "half benched glyph" (if that is the correct term?) so cth, but just ct for example
"d" might work like "y", they are very baked into one another in the text (6819 "dy" matches)

So that is where I am currently, it works for a good portion of the text with lots of previous "freedoms" removed
Next steps are clearing up how "d" works, and "e+gallows"


So if we look at Koens "How many of these are valid Voynichese words?"

tir ty aa na ichin iini tan chom qa aiina ar aaak iini qaiiik latir
dat qa dan qala tana maqar tyny ony kaiinaiin dani tani iti tom eeiti
taniiin lar dana mi daqaea naqan danin tyraqa saiinan okqyiiyny nadik
lich diiinqi donra an iiny iinidin dy ta otka tiras onea iinydyn eeysak
laqi ona iti okqy tam kyklan tani totka eeari aiiar tika onyn idkitoryiin
chynmy tikitoryiin eearchamkan arkim takmi lami miiin lat tadma kam qoiiry
tat okaram tytym tyrata tiri kanriiin okan daniriiin aqaiin okaeam okyr
tarqiiin qaiinam lichtir tikiinitoryiin alanqiiin naraqa tynyn tis
kadammyrkar ika oraqa iinarlata

Invalid (43)
tir - k\k
ichin - \cc\\
latir - k>k\k
kaiinaiin - k>\\\>\\\
iti - \k\
eeiti - cc\k\
daqaea - k>c>c>
tyraqa - kxk>c>
saiinan - cc>\\\>\
nadik - \>k\k
lich - k\cc
diiinqi - k\\\\c\
donra - k>\k>
iinidin - \\\\k\\
otka - >kk>
tiras - k\k>cc
onea - >\c>
laqi - k>c\
iti - \k\
kyklan - kxkk>\
totka - k>kk>
eeari - cc>k\
aiiar - >\\>k
tika - k\k>
idkitoryiin - \kk\k>kx\\\
tikitoryiin - k\k\k>kx\\\
eearchamkan - cc>kcc>kk>\
arkim - >kk\\
takmi - k>k\\
tadma - k>k\>
tiri - k\k\
kanriiin - k>\k\\\\
daniriiin - k>\\k\\\\
okaeam - >k>c>\
tarqiiin - k>kc\\\\
qaiinam - c>\\\>\
lichtir - k\cck\k
tikiinitoryiin - k\k\\\\k>kx\\\
alanqiiin - >k>\c\\\\
naraqa - \>k>c>
tis - k\cc
oraqa - >k>c>
iinarlata - \\\>kK>k>


Valid (57)
ty - kx
aa - >>
na - \>
inni - \\\\
tan - k>\
chom - cc>\
qa - c>
aiina - >\\\>
ar - >k
aaak - >>>k
iini - \\\\
qaiiik - c>\\\k
dat - k>k
qa - c>
dan - k>\
qala - c>k>
tana - k>\>
maqar - \>c>k
tyny - kx\x
ony - >\x
dani - k>\\
tani - k>\\
tom - k>\
taniin - k>\\\\
lar - k>k
dana - k>\>
mi - \\
naqan - \>c>\
danin - k>\\\
okqyiiyny - >kcx\\x\x
an - >\
iiny - \\\x
dy - kx
ta - k>
iinydyn - \\\xkx\           
eeysak - ccxcc>k
ona - >\>
okqy - >kcx
tam - k>\
tani - k>\\   
onyn - >\x\
chynmy - ccx\\x
lami - k>\\
miiin - \\\\\
lat - k>k
kam - k>\
qoiiry - c>\\kx
tat - k>k
okaram - >k>k>\
tytym - kxkx\
tyrata - kxk>k>
okan - >k>\
aqaiin - >c>\\\
okyr - >kxk
tynyn - kx\x\       
kadammyrkar - k>k>\\xk>k     
ika - \k>

It might be interesting to use something like this to determine how likely uncertain spaces are to actually be spaces. Or see if there is any system in attested but "deviating" words. (For example: "something weird may happen after glyph A or B".)

Something feels so very off about seeing kadammyrkar and okqyiiyny as valid Voynichese words though

haha yeah! I think the "slot" system could be something with potential to add rules/structure to this system eventually. 

I think eventually it will be good for identifying unsure spaces, glyphs and weird strings of glyphs. Currier B is going to be a huge hurdle though, well not even a hurdle.. it just doesn't work! Which is interesting.. but I'm ignoring it for the time being, tackle it after A is as good as it can be.

Another quick example
You are not allowed to view links. Register or Login to view.
k>ccc>k (kocheor) cc>k xkcccx kccc>k cc>kcc cc>k>\\\ xk>\\\\ k>\\\
kcc>k cccx cc>k >k>\\\ k>\\ c>kc>k cc>kcc xcc>k>kkc (ychopordg)
c>kcc> xk>k k>\\\ k>\\\ >kcc>k k>\\k c> k>kcc>k k>
c>k>k ccccccc >k>k xk>\\\ ccc>k K>\\\ c>kc>k >kccx k>\\\
ccc>kcc>k cc>k ccc>k>k >kcc>k k>\\\ k>k cc>k cc>k>\\\
>k>k cc>k k>\k>kx

Non-conforming words

kocheor - written "k o cheor" - Only 2 exact cases of "ko". 13 cases of "k". 82 cases of "o". 100 cases of "cheor" - Likely "k o cheor".
ychopordg - the "rd" combo doesn't work with the system, a lot of issues seem to revolve around "d" currently

f1v

kcccx cc>k>\\\ >k >kkcccx (oltchey) cc>k c>kc>k >\
xkccx cc>k >k >ccx kcc> kk>kx (lkody) >k>k>k cc>kx
k> c>kcx c>kc> c>kcx kkcccccx (dksheey) c>kcx k>kcc>kx k>k
k>k cc>kc> k>\k k>\ ccc>cccx (sochey) cc>k>kx

k>k>x ccc>k k>\k c>kc>>k k>k cccx k>kx >k>>\\\ ccc>cccx (shoshy)
cc>kx cc>k c>kc>k ccc>k >k>k k>kcccx cc>k> k>k ccx c>kcx
c> >k cc>kcccc(*) ccc>k k>k c>kccx xk>k k>k k>k> xk>k k> kcc>kx
>k>k ccc>k k>k kcccx cc>k kx xx>k c>kc>k cc>kx cc>k k>\\\
ccc>k >k>k cc>k k>k kx cc>k kcc>k >kcc> k>k ccc>kx
k>>k cc>kcccx k>k cc>kx kcc>kx k>k cc>k>k




Non-conforming words 

oltchey - Likely "o l tchey" looking at spacing
lkody - Spacing looks like "l k o d y" - Not sure what to make of it
dksheey - Odd spacing again, looks like "d k sheey"
sochey - likely "so chey"

shoshy - Likely "sho shy" squeezed up for line end


* P2, L3, W3 is noted as "choeee", however it is "chores", the line after "o" is not a curve and the top flick which joins to the "l" above is faded.
The "s" on the end is clearly "s" and a mistake in the transcription. I have included it as "chores" not "choeee"


Interesting that the non-conforming words are clustered up and have a little more unclear spacing than words around them

I have finished mapping and calculating on the first 10 herbal pages of (20 total);

(Rene Transcription)

Currier A, Hand 1 
Low score - 89.5%
High score -  97.5%
Average - 92.73%

Currier B, Hand 2
Low score - 46.2%
High score - 74.7%
Average - 63.38%

I need to process all the non-conformances, but initial review seems to show "ed" being the main culprit followed by "eos".

A tiny amount of progress

"e" acting as ">" rather than "c" has a huge positive impact on Currier B scores and a negligible impact on Currier A scores. 

It could be that "o" is a pretty universal "switch"
od (2295)
or (2785)
ol (5581)
ok (6097)
ot (3866)
op (578)
of (158)


"a" is fairly specialised 
ad (47)
ar (3192)
al (3032)
ak (37)
at (13)
ap (6)
af (3)


and "e" picks up the bits "a" seems to not like very much, while avoiding pairings "a" likes
ed (5004)
er (32)
el (12)
ek (491)
et (214)
ep (80)
ef (48)


"d" is still problematic. I wouldn't be the first person by a long way to suggest "d" could share a function with "y", but until every possible avenue has been exhausted I don't want to label it "X" (essentially "Null"). I actually don't think "y" is "Null" either, I have some ideas but for the stage I'm at with making this right now the ideas are too complicated. One that I have is that "y" is a word start and end form of "a".

Tiltman (1967) - From You are not allowed to view links. Register or Login to view.
y occurs quite frequently as the initial symbol of a line followed immediately by a combination of symbols which seem to be happy without it in any part of a line away from the beginning. Otherwise it occurs chiefly before spaces very frequently preceded immediately by d. Hence my belief that these two have some separative or conjunctive function. (I have to admit, however, that y also seems sometimes to take the place of o before k or t (though rarely, if ever, after q); this is particularly noticeable in some of the captions to illustrations in the astronomical section of the manuscript - these most commonly begin ok (of) or ot (op) and it is here that we occasionally see yk (yf) or yt (yp).

The basic idea I have is why have another "o", would "a" not make more sense as we do not see "a" starting or ending words (outside of ain/aiin), but is abundant mid-word. also "y" looks like "a" with a tail. Its not a great reason but.. it does. If I wanted to write "ok" that's ok (sorry) but ak is not. If I need to start "a" maybe I have to use "y" because some rigid system I'm using requires it. Maybe the same rule says I must do this at the end too. Just an idea for meddling with stuff at a much higher level once a base ruleset seems, ok (sorry)

I'm not sure why this took me this long to notice, but I think there is a good case (if you think CLS has any merit) that EVA: L, is the line equivalent of EVA: Y. 
This would only leave EVA: Q without a pairing, as it doesn't look to be made up from "c" or "\", but "|", maybe it belongs with other glyphs made up from this (gallows). 

A thought I had, but no idea how much validity it has, is that EVA: q might be attached often to EVA: o (qo) because it is the double looped version, like k>t, f>p.. q> qo?

Anyway, I think the thought process of Y and L being curve/line equivalents and Q not being part of the curve/line groupings was fairly logical and maybe a step forward.