- Options
- Login
- Desktop Style
Options
Decipherment Attempt: A ~1425 Alchemical Manual in Latin Scribal Shorthand
- Decipherment Attempt: A ~1425 Alchemical Manual in Latin Scribal Shorthand
-
Decipherment Attempt: A ~1425 Alchemical Manual in Latin Scribal Shorthand
jredder > 3 hours ago
Hey everyone, what's the happyhaps?
I'm new here, the only other post I made was an introduction post. I'm not an academic, no affiliation with anywhere. I am an IELTS and TEFL certified ESL teacher living abroad (abroad being not my home country lmao).
I don't like mysteries. They confuse and frighten me. A door that never opens is just a wall, and a wall where a door should be is useless. To me, the Voynich Manuscript was a door where a wall should be.
The framework I developed and I are proposing that the Voynich Manuscript is a highly detailed alchemical workbook written in an idiosyncratic scribal shorthand Latin produced by a northern Italian alchemist around 1425 CE.
The author, very likely a Padua graduate from the faculty of physica, has this book full of stuff mostly to keep mercenaries alive. The author was almost certainly a Holy Roman Empire loyalist working to keep the HRE condottieri alive during the conflicts in Northern Italy between the HRE and the Papal authorities.
I know Latin has been proposed before, and every attempt to tie the two together has been beaten to within an inch of its life, but I promise this is at least a bit more robust of a framework than what this community is used to. Greek was proposed nearly to death by people before Ventris eventually proved Linear B was just some old weird Greek.
And I am not only asking you to trust me. I think I am my own worst skeptic, but that is certainly not true, especially not here.
I built tools to test all my work. From beginning to end, you can test it.
The link is You are not allowed to view links. Register or Login to view..
All the code I used is there, you can run it on the website or copy and paste it and run it on your own machines.
There is also a live translator on the website so you can test what I made and see if it makes sense. Either it works or it doesn't lmao.
The framework was done while doing tests on the entire corpus at once mostly, instead of trying to make sense of every page individually. That was too much work and led to nothing.
The shorthand is kind of like an agglutinative abjad system, though there's no good way to describe it efficiently. You'll see when you look at my work.
The book is divided into multiple parts, I'm sure you're aware.
1st section: Herbal section. This section is chimera, or combinations of different parts of plants all frankenstein'd into one plant. It functions as a visual recipe for the potion. They were drawn in the Doctrine of Signatures, which was a common way to draw medicinal plants in medieval Europe, drawing them not by how they look but by what alchemical and chemical principles they had within them. The text on these pages are almost entirely harvesting instructions, distillation purity, what kind of distillation is required, etc. This is just the recipe section of the book.
2nd section: Zodiac section. You see, chemistry and alchemy weren't different things, right? They were the same thing, and they were both governed by the stuff in the sky. You needed to harvest or distill different stuff under different conditions. Harvest peony at night so the woodpeckers don't see you and peck your eyes out (this is a real thing lmaooo), harvest nightshade... at night. Harvest sun flowers when the sun is out, etc. The zodiac part is the reference for harvesting instructions. When certain things happen, moon cycles, etc. This part doesn't have much in the way of instructions, just observations.
3rd section: Balneological section. The weird one, aye? In medieval Europe, there was no periodic table of elements. When you wanted to talk about chemistry and alchemy in visual form, you had... allegory. See, alchemy was divided into three different principles that governed the whole thing. Mercury, Sulphur, and Salt. The mercurial part of alchemy was the raw essence contained within plants and the earth (above ground) that could be extracted and refined. Mercury, volatile, feminine (certainly a choice), and cool. When you see women going through the tubes, it is a very on the nose way to illustrate raw, volatile Mercurial essence going through distillation tubes. The author practiced secular alchemy, pioneered by the OG Maria the Jewess, and we know this because he uses her Balneum Mariae. The balneum mariae used aqua frigida (cold water for cooling the distillation rig) and aqua vitae (the alcohol that you actually distil with). Cold is blue. Life is green. The balneological section is distillation instructions.
4th section: Cosmological section. This one is insane, truly, to look at; however, it's much more simple than the secrets of the universe. It is... topdown equipment setup guide for a Tribikos, or a three alembic distillation rig, which is an invention of, you guessed it, Maria the Jewess.
These are the four main sections for creation, and the last two sections deal mostly with what happens after production, which honestly I am not super interested in explaining here.
As an example of what the parser shows on any given folio, here is a section from 13r:
foli--em radic--em cohobatum Sol-foli--ae
Luna-cum recipe-Luna-cum radic--am canal--us
Leaf, root, cohobation, solar leaf, lunar conditional harvest command, root accusative, vessel noun, etc.
The Tria Prima, which is the governing force behind Southern Alpine alchemy, is encoded in the grammar. Solar markers on solar governed plants, lunar on lunar, and none on salt governed, because nobody cared when you harvest or prepared 99% of roots. They were considered shielded, unaffected by what happens above the ground. It's all encoded in the grammar.
The clearest example of this is folio 28r. Before I looked at the illustration, the parser flagged simultaneous solar AND lunar markers on the same lines, which is statistically unusual. Most folios are dominated by one or the other. My prediction was that the illustration would show a plant requiring both celestial conditions simultaneously. I panicked, thinking my theory was unraveling in front of me, but it was just mandrake. This goofy lil' weirdo is the one plant in all of medieval herbology whose harvest protocol explicitly requires both solar and lunar timing at once. I did not know that before I looked at the picture. I didn't know much of anything before I looked at anything.
As a second example, here's what the parser produces on Folio 70r2, the April zodiac page:
The 2nd ring reads: Solis... cohobatum+magna_mutatio... canal... cohobatum+mensis... inde... cucurbit... foli... Lunam... Solem
That's: Sun's conditions... cohobated charge at Great Mutation scale... vessel... monthly cycle... thereupon... flask... leaf material... Moon... Sun
The token cccc appears exactly once in the entire 40,702 token corpus. On this page. In the astronomical c-series, repetition marks temporal scale: c = day, cc = month, ccc = year. cccc = the Great Mutation. I didn't know what the Great Mutation was when the parser flagged it, which again made me panic because the primary distillation token should only be able to max out at ccc, triple distillation, because... tribikos. The Jupiter-Saturn water triplicity conjunction of 1425 CE, the single most significant astrological event of that generation, falls in March-April 1425. The carbon dating window for the manuscript is 1404-1438. The cccc token sits on the April page. The author was writing during or immediately after the event and marked it in real time. That narrows the composition window from 34 years to roughly 1425-1430.
I did two Monte Carlo permutation tests (in the live repo on my website)
The first one shuffled the semantic assignments, taking all the Latin mappings and randomly reassigns which token gets which Latin value, then runs the parser 1000 times with randomized meanings. If my framework were just lucky or memorizing stuff, you'd expect randomized assignments to score similarly. But they do not! The maximum random coverage it could get across all 1000 runs was 28.02% with roots and suffixes both, where mine got up to 52.1% with my framework.
The second one scrambles the actual characters inside every token in the book. This destroys the internal structure of the text, and then tests the grammar against the fake corpus, which is now all noise. If the parser was just exploiting pure character frequency patterns rather than real structure, it would still score pretty well on random text. It could only do 14.21% in 1000 runs, while my framework did 41.62%.
This is me trying to attack my framework from all directions. One asking "are the right words being assigned to the right meanings" and the other asking "is there actually structure here to find and have I found it?"
They both returned p values of... 0.0000.
Papers are on Zenodo, go take a look:
A Computational Decipherment and Explanation of the Voynich Manuscript - You are not allowed to view links. Register or Login to view.
Tribikos Confirmation, 52.1% Corpus Coverage, and a Complete Translation of Folio 70r2nd_ring - You are not allowed to view links. Register or Login to view.
Tear it apart lads, and as always...
Thank you for taking the time to look at my work. Truly, this community is a blast.
If you have any questions, feel free to message me. I'll be around to talk about it. -
RE: Decipherment Attempt: A ~1425 Alchemical Manual in Latin Scribal Shorthand
rikforto > 3 hours ago
(3 hours ago)jredder Wrote: You are not allowed to view links. Register or Login to view.The token cccc appears exactly once in the entire 40,702 token corpus.
Do I correctly take this to be "eeee"? Which transcript are you using? The one at voynichese.com, great for checking for LLM hallucinations and strange claims alike, has 14 instances of eeee, including the one mentioned here -
RE: Decipherment Attempt: A ~1425 Alchemical Manual in Latin Scribal Shorthand
tavie > 2 hours ago
Your Voynich files don't seem to be there at the link.
Your Linear A stuff is making me think of LLM assisted work. Given Rikforto's question, I'd like to know what AI involvement there was here. -
RE: Decipherment Attempt: A ~1425 Alchemical Manual in Latin Scribal Shorthand
nablator > 1 hour ago
(3 hours ago)rikforto Wrote: You are not allowed to view links. Register or Login to view.Which transcript are you using?
Its V101. As usual AI is unable to understand that V101 is not EVA.
Quote:The transcription corpus used throughout is the standard Voynich interlinear transcription (voyn_101.txt), comprising 40,702 tokens across all manuscript sections (Zandbergen, 2024). Signs are referenced using standard EVA transliteration conventions (Guy, 1991).
A Computational Decipherment and Explanation of the Voynich Manuscript.docx -
RE: Decipherment Attempt: A ~1425 Alchemical Manual in Latin Scribal Shorthand
nablator > 1 hour ago
