The Voynich Ninja

The basic meaning of the word "crib" is a child's bed. But in cryptology it has another meaning.

A crib is an encoding word where we suspect it's meaning or some plain word or expression that we suspect to appear in the ciphered text.
See for example: You are not allowed to view links. Register or Login to view.

Cribs can help a lot in decoding of ciphers and unknown scripts. For example:

- many WWII German messages were cracked because they contained predictable phrases like "Heil Hitler" or "wetter" (weather)
- ancient texts like hieroglyphs contained cartouches ( You are not allowed to view links. Register or Login to view.) , names of kings taken in a border which helped a lot
- Linear B was cracked by Michael Ventris because he correctly assumed that it will contain names of some towns like Knossos
and so on

Cribs are the weakest parts of the ciphers. Unfortunately they aren't fully avoidable because of the nature of human language. You can fight with them but you
won't eradicate them all.

The weird thing about Voynichese it that it doesn't have any obvious cribs and is different in this from other manuscripts.
Would you agree?

For example in herbals it was often very easy to identify the name of the plant. Have a look at the example:
[attachment=13682]

I don't speak medieval Italian. Without googling I don't know what "oriola" means. But I can guess from the text layout it is the name of the plant. If the layout was the same and instead of "oriola" we had "47XAS3D45" then I would suspect that "47XAS3D45" is the name of the plant.

By the way googling doesn't help that much in that case. "Oriola" word isn't used anymore, it means simply "the golden one" and was used for several plants. But it is a problem of historical botany and not cryptology.

Voynich Manuscript doesn't have good candidates for plants names. First words don't really work. See: You are not allowed to view links. Register or Login to view.

In the same way astronomy part should be full of Zodiac sign names. We should have the words for Sun and Moon. But again we don't have good candidates.

Was the scribe consciously avoiding cribs? He gave labels to some nymphs but never gave an label to any obvious thing like the Sun or the Moon. And in other manuscripts scribes often gave labels to obvious things. When they draw a dog they often signed it "dog"   But the scribe never does it in Voynich Manuscript.

So why it doesn't have cribs?

If it is a one to many cipher, where a plaintext word can have many possible encodings, there still can be Zodiac signs and planet names all over the astrological section, but they would be different most of the time.
On the other hand, if we treat the manuscript as a faithful rendering of an unknown language in an invented script, this lack of cribs seems to require some explanation.

(27-01-2026, 02:39 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.On the other hand, if we treat the manuscript as a faithful rendering of an unknown language in an invented script, this lack of cribs seems to require some explanation.

Also the lack of repeated sequence longer than 4 words is puzzling. In Latin herbals there are many often repeated formulas of 5+ words. For example in the Circa Instans: "calida est et sicca in secundo gradu", "virtutem habet confortandi ex aromaticitate"...

Quote:If it is a one to many cipher, where a plaintext word can have many possible encodings

Yes, it is a point. Let me develop it further.

Actually when speaking about one to many cipher you may mean two things which are a bit similar and a bit different

1) homophonic cipher
2) polyalphabetic cipher

1) Homophonic cipher assigns many symbols to one plain letter but the symbols always mean the same:


In the example above there are 3 symbols for "O" but "4" always means "O"

I believe we can safely say that Voynich isn't homophonic cipher. These things have much more symbols than regular alphabet and Voynichese is the opposite - it has fewer symbols than alphabet if we ignore very rare and unsure stuff.

2) Polyalphabetic cipher on the other hand may use the same amount of symbols or fewer than the alphabet but symbols have different meaning depending on their context. It is actually "many to many" cipher and not "one to many". An example would be Vigenere cipher: You are not allowed to view links. Register or Login to view.

So yes, Voynichese could be polyaphabetic cipher which would be very anachronistic but there were stranger things under the Sun

(27-01-2026, 02:03 PM)Rafal Wrote: You are not allowed to view links. Register or Login to view.The weird thing about Voynichese it that it doesn't have any obvious cribs and is different in this from other manuscripts.

You mean, different from other cipher manuscripts...

'Nuff said.  All the best, --stolfi

And going back to the lack of cribs I can see some other possibilities:

1) The images are red herring. They mislead you. The actual text tells adventures of Robin Hood or speaks about hunting badgers  

It would explain why we don't have candidates for plant names or Zodiac signs. But there are another problems - if the text was meaningful and coded in a simple way then we would have another cribs. And William Friedman would crack it whatever it speaks about.
And let's be honest - Voynich Manuscript is weird but it would be even more weird.

Notice that there should be cribs even if the text is in some exotic language or even in a constructed language.

2) The author consciously made an enormous (and apparently successful) labour of avoiding the cribs.

Very anachronistic, paranoid and weird again.

3) The text is gibberish
Well...   Gibberish manuscript is still very weird but for me it is less weird than other options. I will try to develop these ideas in the future.

(27-01-2026, 03:56 PM)Rafal Wrote: You are not allowed to view links. Register or Login to view.2) Polyalphabetic cipher on the other hand may use the same amount of symbols or fewer than the alphabet but symbols have different meaning depending on their context. It is actually "many to many" cipher and not "one to many". An example would be Vigenere cipher: You are not allowed to view links. Register or Login to view.

I'm not sure I use the proper terminology, but for me "many to many" implies that you cannot reconstruct a unique plaintext, there will always be uncertainty. And "one to many" for me means many possible ciphertexts encoding the same plaintext, but only one possible plaintext corresponding to a ciphertext.

Quote:You mean, different from other cipher manuscripts...

Well, I actually meant different from standard manuscripts. Not in terms of the script but the layout.

[attachment=13683]

[attachment=13684]

Nothing is signed:
[attachment=13685]


Other imagery gives you candidates for some concepts just from the layout. Voynich Manuscript doesn't give you candidates.

It gives labels to some nymphs which doesn't help you at all but never gives an obvious label to a Sun or winter.

Quote:I'm not sure I use the proper terminology, but for me "many to many" implies that you cannot reconstruct a unique plaintext, there will always be uncertainty

I am not sure about the terminology too. "Many to many" was my invention on the fly. The proper name is polyalphabetic cipher: You are not allowed to view links. Register or Login to view.

See this example:
You are not allowed to view links. Register or Login to view.

attacking tonight
ovnlqbpvt hznzeuz

I meant that
- "t" from plain text may become "v" but also "n" in the ciphertext
- "z" in the ciphertext may mean "t" but also "i" in the plain text

Actually in Vigenere any letter may mean any another letter.

(27-01-2026, 04:19 PM)Rafal Wrote: You are not allowed to view links. Register or Login to view.And going back to the lack of cribs I can see some other possibilities:

There is also 

4. The language is not any of your guesses

Again, no cypher is as hard to crack as a language that yo don't know.  By a long shot.  

In those examples you gave, the cribs helped only because the guy knew the underlying language, modulo a millennium or two of evolution, and guessed it right.  You can add Hittite (deciphered thank to a crib word "bread" and the guess that the language was Indo-European). 

But the crib will not help if the underlying language has a very different grammar from the languages you know, and very different ways of expressing concepts.  

Linguists have been studying the Etruscan language for more than a century, but their knowledge of it is still very limited.  In spite of hundreds (if not thousands) of "cribs".  Until recently they had identified the words for the numbers 1 to 10, but were still uncertain about 4 and 6: they knew the two words but did not know which was which.   That was solved only by a statistical analysis of the patterns of numbers in Etruscan dice, and the chance find of a pair of dice where the numbers were written as words rather than dots...

The omission of labels on figures copied from other books is expected if it they were copied only for their decorative value.  Imagine the Author telling the Scribe, "here, in the center of this diagram draw something like this" and gave him a book.  That book may have had labels; but since the Scribe did not know Voynichese, he would not know how to translate them. 

But there are many potential cribs on the VMS.  Every label is one.  We only to not know what it is that they are labeling...

There is a crib on page f68r3: the name for the Pleiades (doary, BEEPed as doaro).  Unfortunately it is not mentioned in the text.  Which is not unexpected, as it probably was irrelevant for whatever astrology is discussed in the book.

All the best, --stolfi