Talk:Voynich manuscript

Cheshire article

https://www.academia.edu/117392357/Astrology_Series_No_4_Manuscript_MS_408

2024, Medieval Manuscripts.

Astrology Series, No. 4. MS 408. Gemini: The Lovers. This paper translates all scripts and image annotations included on the Gemini page from the astrology section of the manuscript. Unlike modern astrology, which uses the symbol of the Roman twins Castor and Pollux, this is the preceding Greek version, which uses the lovers Thouros and Phosphoros. This is because Ischia monastery had been established by Basilian monks when the island was part of Magna Graecia (Greater Greece). Rather than discussing personality traits, the author of the manuscript uses the symbol of lovers as an opportunity to provide sex education for the young women of the court at Castello Aragonese, Ischia. The women had been isolated in the castle for their own protection, while King Alfonso V and his men were away conquering Naples, and their natural urges were beginning to show. So, the Gemini page provides instruction for when the men visited the citadel on leave.

(I am reluctant to just add this to the article. What do editors who regularly work on this article think?) Keith Henson (talk) 07:41, 3 November 2025 (UTC) Keith Henson

The link doesn’t work for me but the Astrology heading is of concern. The keys as to inclusion, or not, will be sourcing, weight and academic peer reviews. KJP1 (talk) 08:51, 3 November 2025 (UTC)

I don't think this (preprint?) is any different than a previous discussion on Cheshire. I don't see any third-party coverage showing relevance. - Aoidh (talk) 11:41, 3 November 2025 (UTC)

Are you for real? Aside of the fact that the dearth of male nude figures leaves the supposed audience's education lacking in certain parts (nudge nudge), this seems to be a fairly absurd back story for the VM. (And I can't access the pdf either) --Syzygy (talk) 20:34, 3 November 2025 (UTC)

That's a bad link, so I can't really comment on what it says.

But ... providing a translations of more and more pages doesn't prove that you have a real translation method. I know this seems like a paradox, and makes people angry, but think about this : Millions of people have gotten messages from Ouija boards.

Most VN translations are based on a "method" that is mostly wishful thinking, and yield results that can't be replicated by others using the same method. This is why it's important not to take people's self-published research at face value.

Speaking of which, it is self published, right? I understand that Academia.edu is a self-publishing platform that doesn't have any form of peer-review for their main archive. Better than peer reviewed research would be secondary sources. Experts in the field who validate Cheshire's amateur contributions.

Anyway, Cheshire is already kind of over-represented in the article. I don't think we need to keep adding to his entry.

ApLundell (talk) 02:19, 4 November 2025 (UTC)

I don't know why the URL failed.

Go to https://lingbuzz.net/

Search for Cheshire, click on the pdf

If that doesn't work, post here and I will try again.

Keith Henson (talk) 05:22, 4 November 2025 (UTC)Keith Henson x

I heard from Gerard Cheshire recently. https://ling.auf.net/lingbuzz/008025

Keith Henson (talk) 17:54, 20 January 2026 (UTC)

The Cheshire translation paper "Astrology No. 4." has now topped 40k downloads. Is this significant enough to note in the article? Keith Henson (talk) 20:35, 6 June 2026 (UTC)

Unless reliable sources have reported that as significant in some way within the context of the article's subject, no. - Aoidh (talk) 02:39, 7 June 2026 (UTC)

I would think 40,000 downloads is a significant number even without a reliable source, but have it your way. Keith Henson (talk) 04:25, 7 June 2026 (UTC)

@Hkhenson - No. Please read Wikipedia:Reliable_sources/Perennial_sources, and about Academia.edu in particular, and the advice above. Tony Holkham (Talk) 08:04, 7 June 2026 (UTC)

This article needs to be completely re-written.

This article is out of date, the manuscript has been translated.

It's a medical text, or an herbalary. It's written in 'Proto-Romance'.

I was hoping to find the source article for this here on wikipedia, but no go. ~2026-21823-39 (talk) 02:13, 14 June 2026 (UTC)

Do you have reliable sources to support what you're saying? --Hammersoft (talk) 02:37, 14 June 2026 (UTC)

It doesn't need to be rewritten: you are referring to Gerald Cheshire's theory, which is already in the article. Unlike people who have spent decades studying ancient manuscripts, Cheshire "worked it all out" in two weeks; so far he has not convinced any real expert that his "theory" is more than an overheated imagination. Imaginatorium (talk) 04:53, 14 June 2026 (UTC)

the manuscript has been translated. It's a medical text, or an herbalary -- Aren't these two contradicting statements, demonstrating that the VM has not been translated? ;-) -- Syzygy (talk) 08:00, 15 June 2026 (UTC)

Not contradictory if the sets overlap. —Antonissimo (talk) 22:25, 15 June 2026 (UTC)

What I meant is, if there's doubt about the contents ("medical text or herbary"), it can't very well have been translated, right? (Not that it mattered much, since nobody seems to accept the Cheshire hypothesis anyway.) Syzygy (talk) 08:16, 16 June 2026 (UTC)

And what I meant is, the (premodern) boundary between such concepts is fuzzy; two readers might look at the same text and label it both ways. —Antonissimo (talk) 05:30, 17 June 2026 (UTC)

the, indicates to me that these are two phrases in apposition, that is the second part is a description of the first part ~2026-31253-16 (talk) 13:22, 8 July 2026 (UTC)

Exemplar

not clear to me. There are 2 exemplars? 2 copies? 2 originals? ~2026-31253-16 (talk) 13:18, 8 July 2026 (UTC)

Rainer Hannig

I suggest that this does not belong here, being just another baseless claim. Here is a paragraph from one of the sources (www.rarebookhub.com):

One thing Dr. Hannig's “translation” has in common with all the previous “translations” is that he has not actually translated the manuscript yet. He says it will take some diligent scholars and lots of work to actually produce a translation. None of the theories has actually enabled us to read the document. That would seem to be a basic requirement before we can conclude that Dr. Hannig or any of his precursors (or those that will inevitably follow) have actually broken the code. The proof is in the translation. Until then, we will skeptically withhold judgment.

There are four references for this section:

1. Newsroom, Archaeology (17 June 2020). "Voynich code cracked". Archaeology Wiki. *** written by Hannig himself
2. "The Voynich Manuscript Code Finally Broken...Again". www.rarebookhub.com. *** Quote above: basically "This is just another
3. Shaw, Garry (18 June 2020). "Has Yale's mysterious Voynich Manuscript finally been deciphered?". The Art Newspaper *** Another critical review. He quotes Hannig: “The character of the script, the pronunciation which one needs to get used to, the peculiarity and the vocabulary of the period will cause a lot of trouble even to a native speaker of [Hebrew].” He does not appear to have produced a mapping from Voynich symbols to the Hebrew alphabet, so it is just some sort of phonetic representation? How can the "pronunciation" be strange? Makes absolutely no sense.
4. "Voynich". Rainer Hannig - Autor und Ägyptologe (in German). *** Hannig's own website

So there are two independent sources: note that they both appeared at the time of Hannig's announcement, which was shortly after the Cheshire version. But how many "At last a solution"s have there been since? Probably dozens... I signed up to some sort of feed from academia.edu [bogus warning: this is not a genuine .edu domain], and there are a number of people working away at translations. By and large they each claim a different language for the document, and almost none of it makes any real sense. I think the WP article could perhaps try to show how vast is the world of "solutions", but I am not sure quite how to source this. Anyway, I think Hannig is out of place. Imaginatorium (talk) 17:28, 13 July 2026 (UTC)

The section is certainly not intended to be a comprehensive list of everybody who's ever claimed to crack it. I'd be fine with removing this one, and any other announcement that failed to gain any traction. Doesn't have to be earth-shattering, but some coverage beyond "Here's another one" would be nice.

Cheshire has at least caught some imaginations in the popular press. Hannig's ideas seem to have been ignored. ApLundell (talk) 01:44, 14 July 2026 (UTC)

In the pdf document on Hannig's web page (in German), there is a mapping of most Hebrew letter to the Voynich symbols (page 61). He also gives detailed transcriptions and translations of the first page of the manuscript, and a few lines from a later section. Most of the texts on the first page seem to be quotes from Jewish or Arabic scholars.

Also, "strange pronounciation" refers to, among others, the hypothesis that the letter "h" was not pronounced in this version of Hebrew (he suggests medieval Ashkenazi or Sephardic) and thus often, but not always, omitted in the text. I would argue for keeping this in the article. ~2026-39632-25 (talk) 22:21, 14 July 2026 (UTC)

He also mentions that "w" and "b" were transcribed by the same symbol (ᶟwˁlˁišᶟ might thus refer to arabic scholar Abolays), and "m" and "n" look indiscernible. ~2026-39632-25 (talk) 22:34, 14 July 2026 (UTC)

Is it significant that the writing in the manuscript appears to be from left to right? Tony Holkham (Talk) 23:19, 14 July 2026 (UTC)

I agree that this doesn't belong in the article. It's not a notable decipherable claim and the sources don't make this WP:DUE. This article is about the manuscript, and this unsubstantiated claim is one of countless with minimal coverage and no particular relevance to the article itself. - Aoidh (talk) 08:27, 15 July 2026 (UTC)

They are probably all worth keeping for now, if only to show what has not resulted in deciphering it. But I would like to see each of the claims reduced to the bare minimum, and their sources retained, which will avoid their re-introduction in future. Tony Holkham (Talk) 08:37, 15 July 2026 (UTC)

They are worth keeping if reliable sources show that it's relevant to this article's subject; the ones currently used here fall far short of that. - Aoidh (talk) 08:41, 15 July 2026 (UTC)

Be bold! Tony Holkham (Talk) 08:44, 15 July 2026 (UTC)

I don't think the article is ONLY about the document; it is also about its reception history (otherwise, we'd need a separate article about that). Notability of any particular decipherment claim does not depend exclusively on its being verified (though it would help enormously); notoriety is also a form of notability. The fact that an incredible number of decipherment claims have been made, including wildly amateurish ones (in fact, you could say they all are - a pro would not "claim" a decipherment unless if was well substantiated), is an important part of the reception history, and I certainly believe a select few should be covered in some detail. How many, and which ones, I don't have much of an opinion about - my point is merely that the invalidity of a claim is NOT an absolute argument to exclude it. In fact, I donøt think it is a disaster to have a short paragraph on each of the two or three most commonly reported claims, even if they have to be replaces every few years. Finding good sources may be a problem, though - but a newspaper article is actually a valid source for the fact that the claim exists and has gotten some traction (while it is not a valid source about the validity of the claim). Nø (talk) 16:39, 15 July 2026 (UTC)

The Voynich Manuscript: Zipf's Law etc

Proposed Shift in Methodology: Deconstructing Zipf’s Law and EVA-Reliance in Voynich Studies

Hi everyone,

I would like to suggest a fundamental shift in how we evaluate the linguistic validity of the Voynich manuscript, particularly regarding the long-standing defense against the hoax hypothesis based on Zipf's law. [1, 2]

For decades, the standard text on the Wikipedia article has cited the manuscript's compliance with Zipf's law as a core reason for rejecting the hoax or gibberish theories, arguing that the author in the early 15th century could not have anticipated a statistical distribution discovered only in the 20th century. However, this argument rests on an unverified linguistic assumption that has been mathematically disproven. [1, 3]

1. 1. 1. The Statistical Fallacy of Zipf's Law as Proof of Natural Language

As demonstrated by Wentian Li (1992) in his paper "Random Texts Exhibit Zipf's-Law-Like Word Frequency Distribution" (published by the Santa Fe Institute/IEEE), Zipf’s law is not an exclusive property of natural human languages. Li provided a mathematical proof showing that any randomly generated sequence of symbols (Random Typing) that includes a space delimiter and a natural geometric distribution of word lengths (where short words are structurally more probable than long words) will automatically exhibit a power-law distribution matching Zipf's law. [3] Consequently, the text of the Voynich manuscript does not require a complex underlying natural language to meet Zipf's parameters. A mechanical generation tool, a matrix substitution method, or structured gibberish favoring shorter word units would naturally trigger this exact statistical outcome. [2]

1. 1. 2. Methodological Constraints of EVA Transcription

The historical reliance on the European Voynich Alphabet (EVA) has inadvertently locked computational models and AI decoders into a rigid phonetic paradigm. EVA is fundamentally a mechanical keystroke transcription system designed for computer input, not a structural decipherment. By treating every distinct graphical marker as a separate character, EVA artificially inflates word length and generates repeating consonantal strings (e.g., continuous sequences of c and h) that obscure the underlying morphology.

1. 1. 3. Alternative Research Trajectories

To bypass the current impasse, research should shift from natural-language decoding toward the analysis of mechanical creation methods, testing three distinct structural possibilities:

• Ligature Aggregation: Treating recurring graphic clusters (such as EVA's ch or sh) as unified, single-phoneme characters rather than individual letters, which fundamentally re-scales the text's internal statistics.
• Gallows Characters as Non-Consonantal Markers: Analyzing the high-ascender "Gallows" glyphs (EVA f, p, etc.) as syntax markers, positional delimiters, or vowel indicators rather than standard consonants, given their rigid distribution at paragraph and line headers.
• Cardano Grids and Null-Cipher Padding: Testing the text against early cryptographic shielding systems, where a significant portion of the characters serve as padding ("noise") around localized, sparse semantic outputs.

Rigid adherence to established linguistic constraints has systematically narrowed the scope of cryptanalysis in this field. Because the mathematical literature confirms that random text models readily emulate the statistical signatures historically used to validate the Voynich manuscript as a natural language, the "hoax" section of the main article should be revised to reflect modern information-theoretic consensus. [4]

I propose opening a formal discussion to update the text under the "Hoax hypothesis" to remove the definitive claim that Zipf's law rules out artificial text or structured pranks. [2]

• [1]
• [2]
• [3]
• [4]

PastVoyager (talk) 10:01, 28 July 2026 (UTC)

While your argument has merits, Wikipedia is not the place to do research. I suggest you visit the Voynich Ninja (https://www.voynich.ninja/) or similar established communities dedicated to research into the VM and present your ideas there. --Syzygy (talk) 14:32, 28 July 2026 (UTC)

Thank you for the feedback and the recommendation.Just to clarify, my intention is not to introduce original research (NOR) or propose a personal decryption method into the article. The goal is strictly to improve the encyclopedic accuracy of the existing text.The current article states that Zipf's law definitively rules out the hoax hypothesis. However, peer-reviewed literature in information theory and cryptography (such as Li 1992, Miller 1957, and Ferrer-i-Cancho 2002) has mathematically demonstrated that Zipf-like distributions are expected statistical artifacts of random or mechanically structured texts with word-space delimiters.Therefore, keeping the text as it is presents an outdated and scientifically inaccurate consensus. I am only suggesting an objective update to the "Hoax hypothesis" section based on these published, external academic sources. PastVoyager (talk) 07:47, 31 July 2026 (UTC)

We don't cite those sources directly, because they don't mention the topic of this article, but we do cite this paper which cites the Li paper, and makes the same point you're making, specifically in relation to the Voynich Manuscript. ApLundell (talk) 14:41, 31 July 2026 (UTC)

The final paragraph of the "Hoax" section does touch on the fact that some researcher believe that many of the statistical properties that are used to claim the manuscript must encode a language would also appear in hoax text. ApLundell (talk) 06:02, 29 July 2026 (UTC)