The Voynich Ninja - Why and how the text could be Bavarian

Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

(07-07-2026, 03:42 PM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.Let's keep it simple: I don't know.

Ok, let's focus on the running text. I've pulled the counts and percentages of all different combinations of word ending and word starting character that occur in the manuscript at least 30 times. There are 68 different combinations here. Even if we remove S (and treat ch and Sh as the same) we are still looking at 60 different combinations. 40 different combinations occur more than 100 times. How does the space vowels theory explain these? Are we looking at 60 different vowels? Do we need to further reduce the character set? I don't think any European language has 60 different vowels, including diphthongs.

Code:
31%      3452  11.31%  y.q    

81%      2594    8.50%  y.o    

13%      1929    6.32%  y.c    

28%      1571    5.15%  n.o    

22%      1508    4.94%  n.c    

58%      1330    4.36%  r.o    

50%      1198    3.93%  y.d    

19%      1126    3.69%  r.c    

86%      1120    3.67%  l.c    

02%      964    3.16%  l.o    

50%      757    2.48%  y.l    

96%      749    2.45%  y.S    

26%      703    2.30%  r.a    

42%      659    2.16%  n.S    

33%      584    1.91%  r.S    

14%      551    1.81%  l.d    

88%      530    1.74%  l.S    

29%      430    1.41%  y.k    

61%      405    1.33%  l.q    

90%      392    1.28%  n.q    

08%      362    1.19%  y.s    

25%      356    1.17%  n.d    

43%      328    1.07%  y.y    

44%      310    1.02%  y.t    

28%      254    0.83%  y.r    

10%      252    0.83%  s.o    

91%      247    0.81%  r.q    

71%      244    0.80%  n.y    

48%      234    0.77%  n.a    

17%      210    0.69%  r.d    

83%      203    0.67%  r.y    

46%      193    0.63%  l.k    

08%      189    0.62%  s.a    

66%      175    0.57%  l.l    

20%      165    0.54%  s.c    

61%      126    0.41%  o.c    

01%      123    0.40%  l.a    

41%      120    0.39%  y.a    

79%      117    0.38%  l.y    

16%      114    0.37%  l.s    

52%      109    0.36%  o.l    

15%        92    0.30%  d.q    

45%        92    0.30%  l.t    

75%        91    0.30%  y.p    

34%        88    0.29%  d.o    

62%        88    0.29%  o.o    

91%        87    0.29%  o.d    

18%        82    0.27%  o.q    

44%        81    0.27%  s.S    

71%        80    0.26%  m.c    

95%        76    0.25%  m.o    

19%        72    0.24%  n.s    

39%        62    0.20%  r.s    

59%        60    0.20%  d.c    

78%        58    0.19%  l.r    

97%        58    0.19%  o.r    

16%        57    0.19%  o.k    

49%        47    0.15%  r.k    

65%        47    0.15%  r.l    

80%        46    0.15%  s.y    

07%        40    0.13%  n.l    

19%        37    0.12%  o.S    

31%        37    0.12%  s.q    

43%        36    0.12%  y.f    

54%        34    0.11%  o.s    

65%        31    0.10%  m.S    

74%        30    0.10%  d.S    

94%        30    0.10%  y.e

Oh, come on, we know that fewer than 7 glyphs and families account for over 90% of all spaces.

y = vowels
maybe o = vowels

aiin
aiir
ar/or
al/ol

(m mainly appears at the end of a line and serves as an end marker, to show where and how the flow ends.)

"s" could still stand for “and.”

These families are there to solve other problems that naturally arise in the flow of text. Some are splits of consonantcluster wich are too long, others are typical endings followed by a consonantcluster starting new word, and some are vowels followed by vowels.

This all must also be encoded, and it roughly corresponds to the range they cover.

I’m still working on deciphering this more precisely, which, as I said, isn’t easy due to the spelling anarchy of this century.

(07-07-2026, 08:48 PM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.Oh, come on, we know that fewer than 7 glyphs and families account for over 90% of all spaces.

I thought the scheme works with glyph combinations across spaces? In which case, 41 different glyph combinations cover 90% of all hard spaces (dot spaces). It's the leftmost column of my table in the previous post.

(07-07-2026, 08:48 PM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.y = vowels
maybe o = vowels

...

These families are there to solve other problems that naturally arise in the flow of text. Some are splits of consonantcluster wich are too long, others are typical endings followed by a consonantcluster starting new word, and some are vowels followed by vowels.

This all must also be encoded, and it roughly corresponds to the range they cover.

It's not hard to approximate anything with anything if we add just enough rules. The more rules one has to add to make it work, the less plausible it gets. It still could be right, if it deciphers to something sensible, but without a decoding demonstration it doesn't look very promising to me so far.

But every cipher must also have rules; that’s a circular argument that leads in the wrong direction, and by no means constitutes a counterargument. And so far, there are extremely few and general rules—far from covering every specific case with a new, specific rule. That’s the difference.

(07-07-2026, 09:04 PM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.But every cipher must also have rules; that’s a circular argument that leads in the wrong direction, and by no means constitutes a counterargument. And so far, there are extremely few and general rules—far from covering every specific case with a new, specific rule. That’s the difference.

In general, given the results of your work and the conclusions you have reached so far, I can assume that the cipher was flexible, and the author did not follow any strict rules.

(07-07-2026, 09:04 PM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.But every cipher must also have rules; that’s a circular argument that leads in the wrong direction, and by no means constitutes a counterargument. And so far, there are extremely few and general rules—far from covering every specific case with a new, specific rule. That’s the difference.

I don't think I provided a counterargument, I was talking about plausibility, which is a subjective quality. If I find the method and the set of rules implausible, it's implausible for me, no proof of this is needed.

I'm not sure there is anything to argue about here at this stage, either this method leads to a deciphering, in which case this is huge. Or it doesn't, in which case this is just another attempt to invent a cipher that resembles Voynichese, of which there have been many.

I definitely agree. But I'm not trying to “invent” a cipher; rather, I'm trying to interpret the given information differently. I think that's a crucial difference. Wink

I think I need to clarify what I meant in my last post, because now that I reread it, it sounds as if I'm arguing that no progress is possible without a deciphering. What I meant to say is that if a method provides a sensible deciphering of a large part of the text in a reproducible manner, then it doesn't matter if it can't explain some of the features and peculiarities of the manuscript. In fact, nothing trumps a deciphering that is reproducible and successful by consensus, and it doesn't matter if LAAFU or word length distribution or anything else would turn up a hard to explain emergent property due to some complex interplay of the encoding scheme and the structure of the plaintext, that doesn't follow from the encoding scheme alone.

On the other hand, I think it is still very valuable to have a scheme that seems to align well with many known properties of Voynichese, even if no deciphering is achieved. However, specifically for the space vowels no statistical study has been done yet and I don't see yet now it could reproduce the statistics of the manuscript.

(07-07-2026, 09:12 PM)ololololo Wrote: You are not allowed to view links. Register or Login to view.In general, given the results of your work and the conclusions you have reached so far, I can assume that the cipher was flexible, and the author did not follow any strict rules.

Sorry, I hadn't replied to that yet.

No, so far there are only two rules, but there will certainly need to be more. But even then, I'll do everything I can to avoid using “free” rules as much as possible.

(07-07-2026, 10:07 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.On the other hand, I think it is still very valuable to have a scheme that seems to align well with many known properties of Voynichese ...

I agree with that, too.

And I think it’s clear that this VBM is NOT a “I translated three words, so I’ve solved the VMS” approach.

It’s a very structural approach—I took the peculiarities of the VMS (those strange spaces) and developed a model (!) from them.

Then I realized that this model, with just two rules (delete all spaces; bigrams spanning a space are vowels), already explains some of the statistical oddities of the VMS.

I think I’ve presented this here clearly and scientifically, right down to a possible rationale for extreme word repetition and this strange long-range effect of glyphs.
I’ve also shown other structurea that results from this.

So there’s nothing read into this here—no crazy assumptions, no eisegesis—but rather, it’s simply presented clearly and in a way that anyone can verify.

What it is: It’s a simple model that explains much more than any other simple models I’ve encountered.

And that’s what matters to me for now: to make it clear that such a model can support the idea that the VMS might be based on a completely normal language (note the subjunctive). No more, but also no less.

And of course I’m tinkering with generators, and of course there are statistical features that I can’t reproduce just yet. I’m working on that. But even the Naibbe cipher, strictly speaking, was only near by VMS - and that was based on many more necessary assumptions.

However, since I get the feeling that, because of the length of this thread, many people don't actually know exactly what VBM is, I've created a PDF that explains the current status fairly clearly. I'll update it with any new information and keep posting it here in this thread. That way, no one will have to search to find out what's current and what isn't.

Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36