I recently came across the manuscript: Florence. Biblioteca Medicea Laurenziana, Amiat.1 (viewable You are not allowed to view links. Register or Login to view.), that has an intersting folio. It is 2v-3r, and it looks like this:
Visually, this has nothing to do with the VMS, but I found the concept interesting. It is a top-down view of a palace/bathhouse (can someone confirm this?). If we assume that some circular illustrations in the VMS depict also top-down views of places, such as a bathhouse ("hammams" were already proposed many times for explaining the Rosettes section, for instance), then maybe this is a unique parallel to that kind of imagery.
Recall the "top-down" illustrations from the VMS, such as this one:
And now examine the features of the first picture. All around the edges, we see columns:
And inside, some kind of altar or table with jars on it:
If the VMS author(s) indeed intended to replicate some sort of top-down architecture, then this might somehow provide a clue to it. Can someone more knowledgabe help to translate what this is in the original MS? What are thery trying to depict here?
Having watched the Youtube videos of Voynich Talk, I understand there are only a limited numbers of glyphs in the VMS, certain glyphs appear only at the beginnings or ends of words, some glyphs strongly predict which glyph may follow them, and repeated word families occur throughout the text so a simple cypher letter to letter of a natural language does not match.. Being based in China and very familiar with Chinese characters, it made me think of the following hypothesis: may be the individual glyphs do not represent sounds or letters in the normal alphabetic sense. Rather, they would function more like the strokes used to build characters: meaningless on their own phonetically, but combined according to structural rules to create distinct semantic units. The unit would therefore be the whole “word,” not the individual glyph. This could be in turn used to encode a natural language such as Latin, Italian, German, or Greek encoded in these symbols (as if using some form of ideograms to encode an existing language). Similar-looking words would not necessarily have related pronunciations, but could represent entirely different concepts generated through a a combinatorial system.
What do you think? This could help explain why the manuscript statistically resembles natural language in some respects while not resembling in other aspects?
It would not exactly the same as but a bit like the so-called Wubi typing method in Chinese where each letter refer to a shape or radical combining a set of letter helps to type in characters. Except here there would be an agreed method to assign each word to a character. The weak point of this theory is if it’s highly random it requires XV century copyists to master a whole system which is as complex as Chinese just to produce a single manuscript, that sounds a bit extreme knowing that learning such system would take many years especially in the absence of other texts to learn from...
I don't know if this s new or not. I found some similarities between the voynich letters and a late gothic alphabet. The alphabet can be found on
You are not allowed to view links. Register or Login to view. (St. Gallen, 1438)
I am always looking for ways to find new examples of early 15th century ciphers. As those who are familiar with my online OneDrive archive know I have collected many images of examples of those and there are still some that I haven’t uploaded to my OneDrive. However, I continue to try to increase my collection. So, I have been experimenting with tools like Gemini and ChatGPT to see if they can help generate any new leads. I haven't found anything definitely interesting yet, although they have given me food for thought. However, I wonder if anyone else has experience doing this kind of research.
In this new paper (You are not allowed to view links. Register or Login to view.), companion to the testable signatures paper we recently shared on this forum, we test whether Currier's idea of two distinct "languages" (A and B) in the Voynich manuscript holds up under modern statistical scrutiny. We do this in two complementary ways, using character-pair ratios (like how often 'd' appears versus 'l' on a given page):
First, a generative (unsupervised) model looks at the raw character counts with no knowledge of Currier's labels and asks: how many distinct groups does the data itself support? It independently selects two groups and assigns pages in a way that substantially overlaps with Currier's A/B split.
Second, and perhaps more importantly, a predictive model tests whether knowing a page's A/B label actually lets you forecast its character statistics on unseen pages. The result: it predicts held-out page labels at 89.2% accuracy in character-pair ratios on text the model has never seen.
The A/B distinction is not just a pattern Currier saw: a model rediscovers it blind, and it survives predictive cross-validation.
The A/B label is the dominant axis of variation, but it only explains about 29% of inter-page variance. There's a lot of structure left to account for!
The dataset and methodology are going to be disclosed in two steps: first to a closed group of specialists, then made generally available later on this year.
We present a decoding framework for the Voynich Manuscript. This research provides a repeatable grammatical model, a lookup architecture, a falsification standard, and a translation protocol. The goal is not simply to interpret the manuscript, but to make it decodable in a reproducible way by any sufficiently constrained intelligence, human or artificial intelligence. We provide SEVERAL repeatable translations.
We have also used the same analysis framework to publicly decode The Kryptos K4, Dorabella Cipher, Zodiac Z13/Z32, Beale Cipher and the Shugborough Inscription as proof of rigor.
NOTE: This decoding framework is part of our larger overall research into complex systems science. We are looking for institutional partners (not individuals) that can verify/collaborate for the purpose of producing a case study of our research methods. If you would like to participate please reach out.
Youtube overview: You are not allowed to view links. Register or Login to view.
Full Article: You are not allowed to view links. Register or Login to view.
Hello, this is my first post here so sorry if it is not in a correct area of topic but I have had this thought brewing for a couple days. I have been looking into European handwriting styles and noticed this in an example of secretary Hand.. many letters look oddly similar to voynichese. I am in no way a professional but I thought some people on this website might be able to give some thought on it. (image attached)
I've been working on a computational approach to the Voynich manuscript for about eight months. I'm posting the paper, the code, and all results data because I think the methodology may be useful to the community, and because the negative results are as important as the positive ones.
What I did: I built a 12,365-entry dictionary using logographic character-level assignments and morphological expansion. Then I tested whether the specific character-to-meaning assignments actually matter — whether shuffling which meaning goes on which character produces worse results. It doesn't. 500 permuted dictionaries produce equal or higher coherence scores. The dictionary is dead. That's the first result.
What I did next: I developed a four-gate protocol — pre-registered GO/NO-GO criteria at each step — to see what, if anything, survives falsification. Using morphological decomposition, distributional clustering (PPMI-SVD on the EVA corpus), and Levenshtein reverse-matching against medieval medical vocabulary, I identified five morphological roots whose families pass all four gates: distributional section-tracking, family cohesion above null, cluster enrichment for phonetically-anchored families, and pharmaceutical-section pull under 10,000-permutation testing with Bonferroni correction. Three roots survive even the most conservative correction.
What the five roots suggest: The manuscript contains pharmaceutical preparation vocabulary — roots phonetically matching medieval terms for water, heat, and plant roots — concentrated in the botanical and recipe sections, co-occurring with plant illustration markers (p = 0.018). The text is domain-segregated (z = −55.95 against shuffled null), not multi-layered. The encoding remains unbroken. I cannot read a single line.
What the paper also contains that may be of interest:
- The four-gate protocol run on Rugg-style Cardan grille text — all gates fail (Appendix D). The pipeline does not pass on procedurally generated text.
- Entropy decomposition: root-sequence H₂ = 5.17 bits vs full-token H₂ = 4.80 bits. Affix stripping recovers +0.37 bits. The entropy anomaly persists at the root level. I have not resolved this and I don't claim to.
- Anchor specificity controls: place-name vocabulary produces zero passing families, common Latin produces one, medical vocabulary produces five (Appendix B).
- An abandoned statistical gate (Gate 4 v1) documented as a methodological error and corrected before examining the redesigned result (§5.4).
- A multi-layer hypothesis test that came back negative — I tested my own theory and it failed.
- Illustration-text analysis showing pharmaceutical vocabulary near plant markers, a separate radial-diagram register in the astrological section, and structurally isolated nymph-figure labels with non-pharmaceutical vocabulary.
Code and data: You are not allowed to view links. Register or Login to view.
Pre-registered protocol: PROJECT_8_PROTOCOL.md in the repository
Draft paper Attached and in Github Repo.
The protocol is designed to be reproducible. Everything runs on the Takahashi IVTFF transcription. If anyone wants to run the pipeline on a different transcription or test it on their own hypothesis dictionary, the code is structured for that.
I'm particularly interested in feedback on two points: whether the grille control adequately addresses the generation-mechanism question, and whether anyone has ideas for further decomposing the root-level entropy anomaly by section. Both seem like productive next steps that I haven't pursued.
Happy to answer questions about any part of the methodology.
Hello folks, still somewhat new here, but I wanted to share that I saw a new game pop up in "early access" (game isn't finished but basically needs funding) The game is called: Scriptorium: Master of Manuscripts.
Now I don't for one second believe it to be any sort of level of accuracy or anything of that nature. It is a game built only for fun afterall. But it seems like a fun little diversion.
Thank you, Cow.
You are not allowed to view links. Register or Login to view.
My discovery statements are too long—and I have too little time—to make a compelling anagram. So I am adopting the modern spin on the scientific discovery anagram: a hash.
These are the three key findings of a paper I am preparing for the Voynich 2026 conference. Over-the-top though it may seem, I am taking this step to establish priority in a public setting.
