The Voynich Ninja

julian, one label from f68r3:
oalcheol - venatricum, from venatricis (huntress)
others are pretty ambiguous, I'll try to list possibilities later.

-JKP-, sorry if I wasn't clear. You never stated that it's Latin (you gave some 15% that VM's language is natural, if I'm not mistaken), but you always stress that the author knew Latin very well.

I had a short look at Yulia May's key, and found only one word for which we have the same translation: oty - omnis. That's not very much...

(17-09-2016, 10:58 AM)farmerjohn Wrote: You are not allowed to view links. Register or Login to view.I had a short look at Yulia May's key, and found only one word for which we have the same translation: oty - omnis. That's not very much...

Hello, farmerjohn!
Actually, oty in my interpretation has a few alternates for now, one of them is "opem" and can be omnem, labem, libum, limum, lupum, lobum. I also have a big problem with short words, as I get too many alternates from them. I suppose they may differ by some details in characters to divide one variant of an interpretation from another. My most favourite examples are "o"s: some under the line, some - with a bar, open-loop etc.), but it is still in the process.
A common thing in our theories is that we consider it enciphered Latin text, partially abjad with abbreviations, substituted by one or a few glyphs of the cipher, in particular, y (at a beginning) - com, cum, con... r - r* are also in my key.

(17-09-2016, 11:22 PM)Searcher Wrote: You are not allowed to view links. Register or Login to view.

(17-09-2016, 10:58 AM)farmerjohn Wrote: You are not allowed to view links. Register or Login to view.I had a short look at Yulia May's key, and found only one word for which we have the same translation: oty - omnis. That's not very much...

Hello, farmerjohn!
Actually, oty in my interpretation has a few alternates for now, one of them is "opem" and can be omnem, labem, libum, limum, lupum, lobum. I also have a big problem with short words, as I get too many alternates from them. I suppose they may differ by some details in characters to divide one variant of an interpretation from another. My most favourite examples are "o"s: some under the line, some - with a bar, open-loop etc.), but it is still in the process.
A common thing in our theories is that we consider it enciphered Latin text, partially abjad with abbreviations, substituted by one or a few glyphs of the cipher, in particular, y (at a beginning) - com, cum, con... r - r* are also in my key.

Hello, Searcher.
Ambiguity is certainly the most unpleasant problem for key like ours. It makes them unfalsifiable - nightmare for real scientist
For your solution I strongly suggest to calculate Latin letter frequencies via your key (which I did via my) and in particular to calculate first letter frequency (which I didn't).
These calculations allowed me to understand that t stands for two m/n's and l also is m/n and sh is pr. I also saw that q is q/k and x is z is absolutely acceptable.
And as an amateur with nanoknowledge in this area I don't have right to suggest something to anybody. But if I had , I would advise every Searcher and Researcher to consider possibility that o is u/v/b and pronounced like [w]. This immediately yields that q is q. Then it follows that qo- is aqua- and with no effort at all we got theory explaining why some qo- words are often in Lang B, but not in Lang A.

(18-09-2016, 10:58 AM)farmerjohn Wrote: You are not allowed to view links. Register or Login to view.Hello, Searcher.
Ambiguity is certainly the most unpleasant problem for key like ours. It makes them unfalsifiable - nightmare for real scientist
For your solution I strongly suggest to calculate Latin letter frequencies via your key (which I did via my) and in particular to calculate first letter frequency (which I didn't).
These calculations allowed me to understand that t stands for two m/n's and l also is m/n and sh is pr. I also saw that q is q/k and x is z is absolutely acceptable.

I must say that calculation of letter frequences is possible with your method and, as I understand, exactly this led you to this theory. In my case, firstly, I need to transform ligatures (as I see this) ckh, cth, cph and cfh into usual spelling: eke, ete, epe, efe (vowel-consonant-vowel: aka-, imi-, apa-, etc. ), but this is not a main problem. I think, e, ee, eee and ch are just any vowel and, possibly, v, j and h. Maybe, h is absent at all. I can substitute ee and ch by e, but I will get a whole result for 4 separate vowels and, possibly, for v, j and h. After all, the biggest problem is with "o", of course. Its frequency doesn't mean anything in my theory(!) as I think it can substitute the vowel o and consonants: n and l with or without a vowel (n*, l*), moreover it can be a prefix in-, hence, sometimes it must be examined as a pair of letters, therefore statistics will say nothing. Of course, real scientists, as you say, consider that I'm wrong with this. Anyway, if my supposition about the nature of the cipher is correct just a little, any analysis of the word frequency won't help, it will be false.
I understand that variety of combinations for only one cipher character gives an impression that this will lead to thousands of word alternates, but, in fact, usually I have a few ones for some words, one or two - for another, as well I have no interpretation for some part of words, in particular, I can't interpret labels   Without doubt, a few alternates are also too much for accurate deciphering, so I'm thinking about this all along.
I have a few suppositions and hope to bring this up to scratch:
1. The author uses "prefix-suffix-(ending)" system. For example, "q" is always means "cum", no matter how it is changed due to the next letter: cum-modus (commodus), cum-tentio (contentio), cum-lectio (collectio); "o" as a prefix always means "in": in-positus (impositus), inperator (imperator), in-mediatus (immediatus). 
2. A vowel and a whole word with it can differ not exactly due to its representing (e, ee, ch), but just as sign of one of a certain variant. For example, we have 4 different, but similar words: okedy, okeedy, okeeedy and okchdy. So difference between e, ee, ee, ch may signify a difference between words. For instance: okedy = negatum, okeedy = locutum, okeeedy = lucidum, okchdy = legatum.

Quote:These calculations allowed me to understand that t stands for two m/n's and l also is m/n

Could I ask you? Did you take into account that the most part of doubled m and n in Latin are formed in conjunction of the prefix "com"+"m-root", "con"+"n-root" and "im"+"m-root", "in"+"n-root"? It is important as you also consider that the VMs  "q" stands for the prefix "com", altough I don't know what you think concerning the "in-".

Quote:I would advise every Searcher and Researcher to consider possibility that o is u/v/b and pronounced like [w]. This immediately yields that q is q. Then it follows that qo- is aqua- and with no effort at all we got theory explaining why some qo- words are often in Lang B, but not in Lang A.

Well, it is logical to assume that there must be words related to the water or some liquid in the section with depictions looking like baths. Yesterday I posted about repetitions and sequence of words in that section in You are not allowed to view links. Register or Login to view.. 
So I take the part of a sentence:
 ... ol   kedy   okedy  qoky  okeedy  qokey  qokedy  okedy  qokeedy  okeedy shedy  qoky...
and using your key I try to decipher this part:
um/un dclus(is) udclus(is) aqu(a)dus(is) udcclus(is) aqu(a)dcus(is) aqu(a)dclus(is) udclus(is) aqu(a)dcclus(is) udcclus(is) prclus(is) aqu(a)dus(is)...
What can you say about this?

Searcher, try to unite Latin letters in groups and calculate frequency for the whole group, not for separate symbols. In your case you may define group "vowels".

It's  great if you get few translations for Voynich words, I often get 50+ possibilities  Life without vowels is so... undetermined.

As for m/n calculations. I considered that y gives 1 m/n, yl gives 2, yt gives 3. In general case, letter doublings were not taken into account, d can stand both for l and for ll, but only 1 piece is added to stats. Yes, we loose a bit in precision, say 1% or even 2%, but that's ok. It's only an argument, not a proof.
But if you get some 10% difference, then something is wrong with the key. That happened to me at one stage: mn share among all consonants should be 18%, but my key gave only 10%. This led me to conclusion that t stays strictly for 2 mn's (it was only conjecture before calculations) and there should be another letter for mn - and this is l

(18-09-2016, 07:47 PM)Searcher: Wrote: You are not allowed to view links. Register or Login to view.So I take the part of a sentence
 ... ol   kedy   okedy  qoky  okeedy  qokey  qokedy  okedy  qokeedy  okeedy shedy  qoky...
and using your key I try to decipher this part:
um/un dclus(is) udclus(is) aqu(a)dus(is) udcclus(is) aqu(a)dcus(is) aqu(a)dclus(is) udclus(is) aqu(a)dcclus(is) udcclus(is) prclus(is) aqu(a)dus(is)...
What can you say about this?

I can say this is nightmare for voynichist.
And the most difficult part is d - it can stand for al and Whitaker's Words say that al is suffix, that means "of a ~, pertaining to a ~, in a condition of ~, in a state of ~;" and which, if I understand correctly, can be inserted in many of Latin words...
Is, for example, qokedy derived from qokey by insertion of al? Or d there is a part of the stem?
Nevertheless, I'll try to give translation for each word separately (using Words):
poiin - piavi (cleanse)
olkedy - benedicalis (bless), let it be one word!
okedy - auditalis (hearing)
qoky - aquadis (watered), this should be aquatis, but as I understand t in the middle can sometimes be d. That's what I You are not allowed to view links. Register or Login to view..
okeedy - judicatilis (judge)
qokey - aquadicus
qokedy - aquadicalis
okedy - auditalis
qokeedy - quidditatilis
okeedy 
shedy - protollis (raise up)
qoky - aquadis

No much sense.

(19-09-2016, 08:09 AM)farmerjohn Wrote: You are not allowed to view links. Register or Login to view.Searcher, try to unite Latin letters in groups and calculate frequency for the whole group, not for separate symbols. In your case you may define group "vowels".

It's  great if you get few translations for Voynich words, I often get 50+ possibilities  Life without vowels is so... undetermined.

As for m/n calculations. I considered that y gives 1 m/n, yl gives 2, yt gives 3. In general case, letter doublings were not taken into account, d can stand both for l and for ll, but only 1 piece is added to stats. Yes, we loose a bit in precision, say 1% or even 2%, but that's ok. It's only an argument, not a proof.
But if you get some 10% difference, then something is wrong with the key. That happened to me at one stage: mn share among all consonants should be 18%, but my key gave only 10%. This led me to conclusion that t stays strictly for 2 mn's (it was only conjecture before calculations) and there should be another letter for mn - and this is l

The problem with your suggestion is that searcher is expanding the Latin abbreviation glyphs into a variety of prefixes/suffixes depending on context. Which means the VMS glyph could be two, three, or four different sets of letters in Latin. How do you calculate the frequency of the letters in a paragraph or manuscript when each word-token can have several different interpretations?

In other words, if you have something like yHeey and the y is an abbreviation that behaves differently whether it's at the beginning or end and could stand for two letters or three (in three or four different combinations each), you would have to do frequency analyses for all the different possibilities. Add the EVA-r, EVA-e and EVA-j letters as mutable context-sensitive abbreviations and the factorials for the permutations and combinations for a whole line would be large and impractical for letter-frequency analysis.


If, in contrast, you assign only one interpretation to each abbreviation (so that letter-frequency calculations are possible), then you not only violate the way abbreviations are normally applied in Latin scripts, but you get nothing but gibberish if you try to translate VMS into the Latin language (or any other language).

(19-09-2016, 09:24 AM)-JKP- Wrote: You are not allowed to view links. Register or Login to view.

(19-09-2016, 08:09 AM)farmerjohn Wrote: You are not allowed to view links. Register or Login to view.Searcher, try to unite Latin letters in groups and calculate frequency for the whole group, not for separate symbols. In your case you may define group "vowels".

It's  great if you get few translations for Voynich words, I often get 50+ possibilities  Life without vowels is so... undetermined.

As for m/n calculations. I considered that y gives 1 m/n, yl gives 2, yt gives 3. In general case, letter doublings were not taken into account, d can stand both for l and for ll, but only 1 piece is added to stats. Yes, we loose a bit in precision, say 1% or even 2%, but that's ok. It's only an argument, not a proof.
But if you get some 10% difference, then something is wrong with the key. That happened to me at one stage: mn share among all consonants should be 18%, but my key gave only 10%. This led me to conclusion that t stays strictly for 2 mn's (it was only conjecture before calculations) and there should be another letter for mn - and this is l

The problem with your suggestion is that searcher is expanding the Latin abbreviation glyphs into a variety of prefixes/suffixes depending on context. Which means the VMS glyph could be two, three, or four different sets of letters in Latin. How do you calculate the frequency of the letters in a paragraph or manuscript when each word-token can have several different interpretations?

In other words, if you have something like yHeey and the y is an abbreviation that behaves differently whether it's at the beginning or end and could stand for two letters or three (in three or four different combinations each), you would have to do frequency analyses for all the different possibilities. Add the EVA-r, EVA-e and EVA-j letters as mutable context-sensitive abbreviations and the factorials for the permutations and combinations for a whole line would be large and impractical for letter-frequency analysis.


If, in contrast, you assign only one interpretation to each abbreviation (so that letter-frequency calculations are possible), then you not only violate the way abbreviations are normally applied in Latin scripts, but you get nothing but gibberish if you try to translate VMS into the Latin language (or any other language).

There are some possibilities to deal with this sort of problem
1. Do not calculate frequencies for certain letters.
2. Calculate frequencies for letter groups (not all VM glyphs are abbreviations, I believe  )
3. Match abbreviations to each possibility with certain probability.  If y is A in 50%, and BC in 50%, so y's frequency is spread among A, B and C accordingly.
4. Ignore it
I't up to author to decide if there is any sense in such calculations. For a kind of substitution cipher the answer is yes, imho

(19-09-2016, 10:23 AM)farmerjohn Wrote: You are not allowed to view links. Register or Login to view....
3. Match abbreviations to each possibility with certain probability.  If y is A in 50%, and BC in 50%, so y's frequency is spread among A, B and C accordingly.

But that's the problem.

It's not only that each abbreviation-letter (and there are many of them) can stand for several different things AND can stand for something different depending on position, but EACH WORD in the line (and the paragraph) can be evaluated differently even if it's written the same (and especially if it's written differently with even one letter difference).

They compound each other. Using "abbreviations with certain probability" doesn't solve the compounding dynamic. We're talking about very big numbers here.

Finally, I have managed to translate several tiny excerpts 
Also the key was slightly updated.
For those who are interested please see the attachment.