I found the following new article about the VMS titled “What We Actually Know.” While the content goes beyond that, I think the basic approach to working with the VMS, as described in part, is also worth reading.

You are not allowed to view links. Register or Login to view.

I think we should collect all German medicinal manuscripts containing instructions for remedies involving charms here and rate them. In equivalence to the month names, it would be interesting if we can pinpoint a certain area or date range.

Provinzialbibliothek Amberg Ms. 77 - Varia opuscula medica scripta (Arzneilehre, Kräuterbuch, Amberger Malerbüchlein)
You are not allowed to view links. Register or Login to view.

Unfortunately the age is only given as 'Mittelalter', and no location.
Provenience: from the Cistercian abbey of Waldsassen in Oberpfalz, Bavaria. In the Holy Roman Empire it was an Imperial Abbey.

Contains german recipe, latin charm, crosses, so nim, gaissmilch, large line-end 'o'. Letter 'g' contains a rightward horizontal stroke.

Hello, 

I wanted to raise some potential issues that I've had with the carbon dating of the VMS. Before I start, I want to make abundantly clear that I accept the dating for the individual leaves of parchment as reported, and that this is NOT an "anti carbon dating" post in any way.

I believe that some of the core assumptions used to come to the final dating range do not necessarily hold up to scrutiny. Before I do so, I want to make clear the following things that I have no issue with: 

- I see no issue in the handling or testing process for any of the individual samples
- I do not disagee with any of the individual dating results for any of the samples
- I believe that the results were correctly calculated after the core assumptions were applied (I ran the numbers myself as well and got pretty much the same range)

---------------------------------------------------------------------------------------------------------------------------------------

With that out of the way, let me explain what I mean. My issues stem from the combination of the invidual dating results into a smaller range using a statistical process. As I mentioned above, I believe that this statistical process was done entirely correctly after the core assumptions were applied. But, what were those assumptions that justified doing so? 

1) The individual leaves of parchment are part of a singular "whole" object, and thus can all be assumed to be samples from that same "object"

What is the justification for that to be assumed as correct?  As far as I can tell, the main reason to accept this is that the entire VMS contains "voynichese". That's pretty much it.  As i've frequently come across while looking through manuscripts, it is common for there to be a seemingly random inserted page, sometimes far older, sometimes far younger within the binding. And often the only discernable way to know that it is dated differently is the content: perhaps the handwriting is different, or the language is different. We have no such luxury with the voynich manuscript, as we cannot understand the content.

Okay, but isn't this part of the reason why samples were taken from different sections? Yes, but there are FAR too few samples to conclusively say that there are no outlier leaves. If the manuscript was indeed previously loose sheets in a different order, it's completely possible that there are some older or younger leaves scattered within the sections. For instance, say that someone began writing the VMS in 1380AD, and completed around 4 pages, but never finished the rest or died. Then, a student or family member continued the work in the same style, with the same writing rules, later in 1420AD. When the book was rebound later, those 4 pages would not have stood out from the rest, and would now be inserted at random within a section. 

So seperately, it is possible that if we were to take 4 more samples from 4 other leaves, we may happen to get completely different results. That is the nature of seperate pieces of parchment, and more importantly that is the nature of those sheets NOT being a single object. Let me present this another way: if 20% of ALL folios were truly from 1350AD, and 80% were from 1420AD, the chance to pick 4 samples from 1420 at random is around 40%. 

2) The VMS was written and produced in a short period of time

I don't understand the evidence for this. How can we possibly know? 

As I mentioned above, there are countless instances of composite manuscripts containing works from different time periods. Quite often, there is no real discernable difference between the quality of the parchment - it is only the content that changes. Sometimes, I come across examples where the text was written at a certain time, and the illumination didn't take place until decades later. Take this example, where the text was written in 1425-1450AD (or maybe slightly earlier), but the illuminations date later, with some dating to around the last quarter of the 15th century: You are not allowed to view links. Register or Login to view.

If you were to take 4 samples from that manuscript, how many would come back saying 1490AD? 

The biggest issue with this assumption is that it effectively stifles research into anything post 1450AD, when it is entirely possible that sections of the VMS were written after 1450AD. We can point at certain aspects that potentially date the VMS, such as the clothing, or the zodiac signs, or the swallowtail merlons. Okay, but that dates those pages, not the entire VMS. It actually doesn't date those pages with certainty either; there are plenty of examples of almost 1-1 copies of illustrations over decades. 

Seperately, this assumption itself implies something: that the works in the VMS must date to the same time period as the parchment. This, I believe for obvious reasons, is not a reasonable standpoint. What is to say the VMS author was not sourcing from an italian manuscript from 1320AD, and seperately sourcing another french manuscript from 1410, and then also sourcing a german manuscript from 1250AD? Sourcing works from 1250AD sounds unlikely, but to do so in 1430AD would be almost exactly like us sourcing from Charles Dicken's Great Expectations, from 1860. Not farfetched in the slightest. 


------------------------------------------------------------------------------

The 14th century f68 peak

A seperate issue that comes forth after doing this statistical process is the dismissal of the huge early-mid 14th century peak in the f68 result. This is effectively brushed aside by aggregating the results of the different pieces of parchment. Is this proper? It seems entirely improper to state that a single piece of parchment has a 42% chance of being dated to 1308-1362, but then to take 3 other seperate samples, state that they are the same object as f68 (when they are not), and then say there's a 95% chance that the whole thing dates to 1404-1435.  

This is basically saying "yeah, theres that huge peak there, but these other pages dont have the peak, so its probably worth ignoring". Why can't one page genuinely date to 1330, and another genuinely date to 1420? 

Again, if we had chosen 4 other pieces of parchment, and each had the same ~50-50 peak distribution as in f68, there would have to be a huge discussion on which peak was more likely. But because we only found one, we dismiss it entirely. It's not right. 

If the reality is genuinely that some pages date to ~1330, and others date to ~1420 (coincidentally at the same levels of atmospheric carbon), we would expect to see peaks in both for both sets of pages. 

--------------------------------------------------------------------------------

Conclusion

Basically, this is what I am making the thread to state: 

- The dating range of 1404-1435AD is based on what I believe are potentially incorrect assumptions. 
- Based on the results, we should probably have a likely working date-range of ~1385-1458AD
- More carbon testing could potentially find more pages similar to f68, with a large peak in the mid 14th century
- To say that the VMS was written in a short period of time, or saying that the VMS is dated fom 1404-1435, even if true, puts unneccesary restrictions on various aspects of research.
- We should be treating works from the 9th-15th century as possible direct inspirations for the content in the VMS, instead of broadly dismissing most as probably not relevant.
- We should not presume that the content cannot have been post-1435AD

I know that making this thread will probably not win me any favours, and i've been sitting on it for a while now debating whether to post. But if anything comes across as an attack on other researchers and their work, please do not take it that way. I don't mean for that to be the case at all, and I have huge respect to those that have done that work.

Spencer Coll. MS. 65 is a 15th century Italian herbal in the Taccuinum Sanitatis tradition. But it is an unusual one. In fact, it might somehow even be related to the VMS!

This herbal is written in Italian and has the peculiar feature of having illustrations of castles and towers with Ghibelline (swallowtail) merlons. In fact, there seem to be a lot of them:

   

   

   

   

   

   

There even seems to be several castles with Ghibelline wall and towers all around (note the first picture below, in which there appear to be bathing people inside):

   

   

Can someone fluent in Italian check which castles these might represent? That could be a clue to the Rosettes section castles of the VMS!

But there are other scenes that might be related to bathing, too:

   

   

Hoow about this weird animal, kind of reminds me of the VMS "armadillo/pangolin":

   

Some herb roots are also voynich-y:

   

   

You can find the MS here: You are not allowed to view links. Register or Login to view.

Could this MS be relevant to the Voynich manuscript in some way?

Hi everyone,

First of all, I am an amateur when it comes to both the Voynich Manuscript and historical linguistics. English is also not my first language, so apologies in advance if I misuse some terminology.

I am not proposing a decipherment, and I am not claiming that Voynichese represents any particular kind of language.

I would like to ask whether the following kind of controlled experiment has already been attempted, and whether there is an obvious methodological problem with it that I am overlooking.

The basic question is:

Quote:If a Herbal plant and a Pharmaceutical plant fragment are matched independently from the illustrations, does the associated text preserve a statistically detectable relationship between the two?

• genuine morphology
  
• phonological or syllabic restrictions
  
• a cipher or another encoding system
  
• a system of abbreviations
  
• an artificial procedure generating word-like forms from a limited set of components
  

• How strongly are its characteristic short glyph sequences represented on the visually corresponding Herbal page?
  
• How frequently do the same sequences occur on other comparable Herbal pages?
  

Quote:Does the exact same word occur in both places?

Quote:Does the visually corresponding Herbal page contain an unusually high concentration of the same relatively rare sub-word elements that occur in this Pharmaceutical label?

Quote:Is the text in this part of f102 statistically more similar to the group of Herbal pages visually associated with its plant fragments than it is to other comparable Herbal pages or groups of pages?

Quote:Does the text contain statistical information about relationships between objects that can be identified independently from the illustrations?

1. Meaningful text.
   If both entries concern the same plant, some lexical, morphological, or otherwise object-specific component might survive in both contexts even if the rest of the text is completely different.
   
2. Artificial or self-referential pseudo-text.
   Some form of artificially generated text would still be possible. However, the generation procedure would apparently have to be more complicated than simply producing locally similar words or drawing from a general "plant vocabulary".
   It would somehow have to preserve object-specific relationships across different parts of the manuscript.
   
3. A common source or exemplar.
   Both the illustrations and some of the associated textual elements might have been copied from a common source.
   In that case, the person producing the Voynich Manuscript would not necessarily even have to understand the repeated textual elements. It would be sufficient to reproduce information that was consistently associated with the same object in the exemplar.
   

Quote:This particular method failed to detect an object-specific textual relationship.

1. Selects Herbal–Pharma correspondences in advance using only the illustrations, without looking at the text when deciding which objects match.
   
2. Tests object-level relationships by comparing individual Pharmaceutical labels with rare or informative sub-word structures on the corresponding Herbal pages, rather than only searching for identical complete words.
   
3. Separately tests the f102 material at the page, panel, or section level against the group of Herbal pages visually represented there.
   
4. Compares the observed similarities against suitable alternative Herbal pages or groups, rather than simply identifying interesting matches after the fact.
   
5. Takes into account the fact that several corresponding Pharmaceutical fragments may occur on the same page or panel and therefore should not necessarily be treated as statistically independent observations.
   

Paragraph initial words primarily, but not always, begin with a plain gallows (K, F, T, P) and almost never a benched gallows - but…

When a paragraph does start with a benched gallows it is most often going to be a split gallows.

Bear with my slight variation of transliteration for the peculiarities of a split gallows.

Not split:
F1r cpho
F65v cphy 
F76v cphdar 
F101r cphoar 

Split:
F8r ctho-cthey 
F8v cthod-soocth 
F30v: cthscthain 
F35r cthoo-r-choly-cthy
F42r stho-ofaiin-cthaihcthy 
F80r: cthcthey 
F86v6: ctholcthy
F90v2 cphdaithy 
F100v: cthdeecthy 
F102r2: cthol-dcthody 

Not quite a bench:
F101r: etolctheol 
F111v: psalar (not split either)

Non-bench split gallows:
F56r: Otchal-chchsty 
F68r1: poeeodcthy (label)
F68r2: opol-chcthy (label)
F95v1: polteedy-ctho
F99v: oteol-socthey 

The other ‘decorative’ paragraph initials where a P extends over one or more words before returning to cross the leg might also wrap those words into one highlighted segment. Of course, we may never know if they mean something at all, but I still like to think there’s more to these markers than just decoration.

I have been considering the possibility that the Voynich manuscript uses some form of lossy encoding.

By “lossy,” I mean that the source text may originally have been written in Latin, but that only part of its linguistic information was preserved in writing. Such a system would assume that the intended reader already possessed enough practical knowledge, contextual information, or interpretive tools to reconstruct the meaning. Where the result might otherwise be ambiguous, the system could provide additional cues to guide the reader toward the correct interpretation.

Shannon’s information theory suggests that natural languages contain a considerable amount of redundancy, sometimes estimated at around 50 percent. Latin, as a highly inflected language, carries much of its grammatical information in word endings. Nouns, adjectives, and verbs are marked according to their roles in the sentence, meaning that some structural information is communicated more than once.

In a practical text, however, much of this information may not be strictly necessary. If a practitioner already knows how to prepare a decoction, grammatical case endings and many function words may add relatively little to the basic operational meaning.

To see how this might work in practice, consider a short apothecary-style recipe in Latin:

"Ad dolorem capitis: Recipe radices hellebori, et tere in mortario. Misce cum aceto et oleo rosarum. Ungue frontem patientis ante somnum."

In ordinary English, the passage means roughly:

"For headache: Take hellebore roots and grind them in a mortar. Mix with vinegar and rose oil. Anoint the patient’s forehead before sleep."

If we reduce the Latin words to their basic dictionary forms, we get something like this:

“ad dolor caput recipere radix helleborus et terere in mortarium miscere cum acetum et oleum rosa unguere frons patiens ante somnus”

We can then remove prepositions such as ad, in, cum, and ante, conjunctions such as et, and predictable instructional verbs such as recipere and miscere. What remains is the raw referential payload:

"dolor caput radix helleborus terere mortarium acetum oleum rosa unguere frons patiens somnus"

The result is an awkward, almost “caveman-style” form of Latin. To a modern linguist looking for conventional morphology and syntax, it appears broken. To a trained fifteenth-century apothecary, however, much of the intended meaning would still be recoverable. The practical context constrains the possible relationships among hellebore roots, a mortar, vinegar, rose oil, headache, and the patient’s forehead.

This is important because practical knowledge does part of the interpretive work that grammar would normally perform. The reader does not have to consider every theoretically possible relationship among the words. Existing knowledge of the craft or some other external knowledge narrows the range of plausible interpretations. The system expects that the receiver has sufficient information to handle the context recovery, transmitted beforehand.

The difference between the information carried by ordinary Latin and the apparently smaller capacity of the encoded text might therefore be explained, at least in part, by removing inflectional endings, function words, predictable instructions, and other contextually recoverable material.

Once Latin is reduced in this way, however, the result no longer looks much like Latin. Ambiguities also inevitably arise. For example, simply listing an ingredient and an action does not always tell the reader what is being acted upon, in what order, or under which conditions.

To preserve meaning without ordinary inflections, the encoding system would therefore need an alternative way of organizing the remaining roots or semantic units. One possibility is a rigid positional framework in which an element’s location determines its function. A position might indicate whether a root refers to an ingredient, an action, a quantity, a body part, a preparation method, or some other category.

In effect, position would take over part of the work normally performed by Latin morphology and syntax. The resulting text would not simply be abbreviated Latin. It would be Latin material reorganized within a different information system, one that relies on shared professional knowledge, predictable context, and positional structure to recover what has been omitted.

Any views or comments on this?

Thank you.

-Pointless

So, some of you may remember my work where I created a 'statistical' generator using what I called a ledger.  You can find it here: You are not allowed to view links. Register or Login to view.

Well, I have discovered why it didn't exactly inspire anyone. My logic was way off.  I have since been improving on my methods and digging deeper into the possibilities of the ledger and I believe I have something that works MUCH better.  Before I bury you in data I'll have to state that I have a LOT of information to impart so it will take several posts to explain all of this.  I'll start with the big news first then the explanations.

Before I go any further, I want to give credit where it is due. This work follows and expands on the work of Torsten Timm and Andreas Schinner.  Their self-citation model proposes that Voynich words were produced by copying existing words and modifying them. This is the foundation of what I have been calling “copy and mutate.”  So if you're hoping for a decoding, TLDR; it's not here.


Positionless ledger

1. Look at the prefix column for the row that contains D.
   
2. In the core midfix column you'll see you can choose A.  You now have D→A
   
3. Now look in the prefix column for the row that contains A.
   
4. Look in the core midfix column and you'll see you can choose I. You now have D→A→I
   
5. Look in the prefix column for the row that contains I. 
   
6. Look in the core midfix column and you'll see that you can also choose I as a midfix. You now have D→A→I→I
   
7. You'd again look in the I prefix row and in the suffix column you can choose N
   
8. You now have the word D→A→I→I→N
   

• 393 core word types
  
• 412 rare word types
  
• 805 total word types
  
• 12,564 manuscript tokens
  

• aiin
  
• daiin
  
• odaiin
  
• chodaiin
  

• aiin
  
• d + aiin
  
• od + aiin
  
• chod + aiin
  

• CH→O 
  
• O→D
  
• D→A 
  
• A→I 
  
• I→I 
  
• I→N
  

• N
  
• I → N
  
• I → I → N
  
• A → I → I → N
  
• D → A → I → I → N
  

• chedy 500 occurrences
  
• lchedy 108 occurrences
  
• olchedy 34 occurrences
  
• dolchedy 3 occurrences
  

• chedy
  
• L + chedy = lchedy
  
• O + lchedy = olchedy
  
• D + olchedy = dolchedy
  

• CH → E → D → Y
  
• L → CH → E → D → Y
  
• O → L → CH → E → D → Y
  
• D → O → L → CH → E → D → Y
  

• D→O 
  
• O→L 
  
• L→CH 
  
• CH→E 
  
• E→D 
  
• D→Y
  

• chedy = C → H → E → D → Y
  
• lchedy = L → C → H → E → D → Y
  
• olchedy = O → L → C → H → E → D → Y
  
• dolchedy = D → O → L → C → H → E → D → Y
  

• chedy H in the first midfix position
  
• lchedy H in the second midfix position
  
• olchedy H in the third midfix position
  
• dolchedy H in the fourth midfix position
  

• chedy = CH → E → D → Y
  
• lchedy = L → CH → E → D → Y
  
• olchedy = O → L → CH → E → D → Y
  
• dolchedy = D → O → L → CH → E → D → Y
  

• D→O 
  
• O→L 
  
• L→CH 
  
• CH→E
  
• E→D 
  
• D→Y
  

Note to new members trying to register. There is currently an issue with confirmation emails not being sent. We are working on this - I will report here when it is fixed.

(22-08-2026, 03:59 PM)pfeaster wrote: You are not allowed to view links. Register or Login to view.But they also report finding that "certain" spaces tend to fall between their study's learned units, while "uncertain" spaces tend to fall within them. That seems potentially more meaningful in terms of supporting the idea of a hierarchy of "strong" and "weak" spaces.