The Voynich Ninja

Here are the assumptions -- that are admittedly far, far from having proof to support them:

1) The VM is a ciphered text, with an actual, widespread natural language plaintext underlying it
2) Every use of EVA "y" within the text is a null (that is, that glyph has has no underlying corresponding character (or characters) in the plaintext that it is represented by) -- basically, someone who is decoding the VM message should IGNORE that character

Given these two assumptions, I wanted to examine Koen's work on removing EVA y (after the bigram alterations).  The result he found was that entropy is signficantly increased (e.g. there is significantly more disorder).

Now if we use the traditional definition of a null -- that is, that symbol in the cipher "does not care what other symbols it sits next to" it is clear that EVA "y" is not behaving how a traditional null is defined.  

I was wondering about the traditional use of nulls at the time of the VM carbon dating.

Does anyone have any insight into the use of nulls in 15th Century ciphers?  That is, are we getting "anachronistic" with some sort of non-traditional null.  By non-traditional null I mean, after every symbol (or pair of symbols) that stand for plaintext letter X, add a null, rather than "random" scattering.

I am finding it hard to imagine the motivation for a"non-traditional" null other than in relation to confounding some sort of frequency analysis, which I understand is a further anachronism.  But if someone can come up with an alternative, I would be very interested in hearing about it.

Well -- I guess I have now answered my own question, as I suppose it could be hypothesized that nulls do not occur at the glyph level, but instead at the whole word level -- and if those null words were highly ordered, removal of a portion (like the "y"s only) could have this effect on entropy, even if the nulls were actually used in a "traditional" sense, as defined above.

In any case, would be interested in any other thoughts along these lines or just some comment on how varied the use of nulls were at the time and thanks for thinking about it!

It was Nick who commented about the supposed expected behavior of nulls. If I understood his comments correctly, we should indeed expect the removal of nulls to decrease entropy. It seems to me though that Nick was thinking more about relatively rare characters as nulls - relative compared to [y]. But I hope he will comment on this himself. That said, as I demonstrated it is extremely easy to decrease entropy by introducing nulls. This is not a complex operation, and it may be something anyone at any time could come up with. 

What I find most interesting - though I have not looked in detail at it yet - is Rene's study about which character has the highest information value in VM words. Combining this knowledge with null-hypotheses (   ) may have interesting results.

I would be inclined to say that normally nulls can be put in any position within a cipher text at the discretion of the author. This does not preclude given null symbols being used only in certain positions, but I can't say that I am aware of any examples of this being done specifically and I have seen a lot of cipher keys from the 15th century.

One thing I would be interested in hearing Nick Pelling's thoughts on is the varied use of Tironian notes within 15th ciphers, it's not something that I have really got my head around. (There is an example of an early 15th cipher key from King Rupert of Germany in the Gonzaga cipher ledger and I guess it is using some kind of shorthand, but I don't understand quite how it works and it would be interesting to.)

The problem with nulls is that there is no such thing as a single "null strategy" for using them. In 1420, an encipherer using basically a MASC to encipher Italian or Latin might (very sensibly) add nulls to the ends of words to try to disguise the (tell-tale) word-endings. A 1450 encipherer (with two or three homophones per plaintext letter) might insert nulls all over to try to disrupt common words (such as "lo" etc) or just to pad out short words (that might otherwise be giveaways).

The idea that EVA '-y' might be a null is therefore an interesting one: but I think its strong affinity for pairing with 'd' (e.g. '-dy') is very hard to explain in those terms. My 2006 suggestion that '-y' probably stands for (truncatio) and '-d-' for (contractio) is still just about the only show in town.

All the same, the right approach to be taking here would be to find enciphered documents from 1350-1450 and see how people actually used nulls. Having nulls in the cipher key is no help at all in understanding cipher practices.

Mark: even though there are a few Tironian-like notae fossil shapes in the Voynich, I don't think there's any real likelihood that these actually function as notae. But I will try to have a look at the Gonzaga cipher ledger to see what's going on there.

I think the "y" variant is more comparable with plural or singular. How exactly he does it is only a guess. Since I know that he also uses German, it is possible that he uses both.

Already seen: ma9, mu9 = ma-us = Maus , mouse

Seen in this light, Eva M9 would simply be mus = muss , must
den mus de......

Thank you, Nick and Aga for the possible “structural” uses of nulls.  They could be used in a targeted way to hide word length as well as tell-tale locations of particular symbols (if the word structure of the underlying plaintext was too predictable (like Italian with too many ending vowels or German with a too obvious way of making plurals)) - and none of this requires a frequency analysis.

Nick, l also agree that the frequency of the EVA -dy pair gives me pause.  It “feels” artificial like EVA qo- to me - but l have a feeling the function of these two very common pairs is unlikely to be the same or mirrors of each other somehow.  But to figure it out we may well have to get “lucky” and pick right functions from the myriad of possible ones for not just one but both of them because of the high frequency of these pairs in the text.

Koen - quite interested as always for future data.

(03-10-2020, 11:00 PM)nickpelling Wrote: You are not allowed to view links. Register or Login to view.All the same, the right approach to be taking here would be to find enciphered documents from 1350-1450 and see how people actually used nulls. Having nulls in the cipher key is no help at all in understanding cipher practices. 

Yes, l agree.  Any thoughts on sources?

Although it does strike me as a possible thesis in historic cryptography - “The Use of Nulls in 15th Century Ciphers”  - although l’m a little concerned you’d have to expand at least into the 16th Crntury to have enough material for a decent degree defense!

(04-10-2020, 12:33 AM)MichelleL11 Wrote: You are not allowed to view links. Register or Login to view.

(03-10-2020, 11:00 PM)nickpelling Wrote: You are not allowed to view links. Register or Login to view.All the same, the right approach to be taking here would be to find enciphered documents from 1350-1450 and see how people actually used nulls. Having nulls in the cipher key is no help at all in understanding cipher practices. 

Yes, l agree.  Any thoughts on sources?

Although it does strike me as a possible thesis in historic cryptography - “The Use of Nulls in 15th Century Ciphers”  - although l’m a little concerned you’d have to expand at least into the 16th Crntury to have enough material for a decent degree defense!

There are quite a few sources. Are you actually undertaking academic research?

(04-10-2020, 12:33 AM)MichelleL11 Wrote: You are not allowed to view links. Register or Login to view.Yes, l agree.  Any thoughts on sources?

I'm not sure. Cipher historians tend to overfocus on cipher ledgers, because they're such a rich (and often surprisingly diverse) source of information (both technical and social), but they (of course) bracket the whole issue of scribal cipher practice (e.g. nulls, but also homophone usage, and mixing plaintext with ciphertext).

I'd need to look back at the sources. Cerioni mentions lots of letters that were decrypted (presumably centuries) later, but I'm not sure these have been examined from the point of view of practice.