(10-06-2026, 11:26 PM)ololololo Wrote: You are not allowed to view links. Register or Login to view. (04-06-2026, 12:10 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.Well, when I wrote my last post, I didn't think about sharing my opinion on this... I think it would be better if I wrote it in a separate post.
During my independent study of the Voynich manuscript (which took about a year), I came to the conclusion that Voynich is most likely a cipher based on numbers. I am not a cryptology expert, but I will try to explain how I came to this conclusion.
Let's start with what I wrote about in the previous post, which is that the text of the Voynich manuscript (or rather, the mechanism of word formation) is very similar to classical substitution: all words consist of a fixed set of characters and bigrams arranged in a certain order. Here you might think that I am reinventing the wheel, having in mind the concepts of "prefix-midfix-suffix" or Stolfa's "crust-mantle-nucleus" model (in fact, this is what some initially thought), however, this is not quite the case. The set of supposed minimal substitution units that I have compiled demonstrates some more dependencies besides the dependencies of position in the word.
I would like to suggest that you look at this from a different angle. If we can easily decompose any word in the manuscript into such minimal units, but at the same time we clearly see that this is not a regular "symbol-symbol" substitution, then we can use regular numbers to help us.
First, I'll add a list of the "letters" I've generated:
1). Single letters: o, d, e*, y, r, l and maybe s,
2). Bigrams: oi, ai(an), or, ol, ar, al, om, am, in, iin, ee (es)*, ch/sh, ir, il, im, qo
3). All gallows and EVA x.
Pay attention to the digrams al-ar and ol-or. Their peculiarity is that in such a combination they repeatedly appear both in words and standing alone. But by swapping the letters (ol-or to lo-ro, al-ar to la-ra), these properties are immediately lost, and the structure of words becomes "more fragmented" (let's take the word for example oralar. It can be decomposed as ol-ar-ar, and as o-la-ra-R. I think you can immediately see the difference between three bigrams and two extra letters around the edges. In addition, there are no words "lara" or "rala" in the manuscript). This remarkable property of bigrams suggests that the letters of the Voynich manuscript are not Latin letters, but numbers.
But if they were Arabic numerals, swapping the digits in a number would result in a different number (for example, 41 becomes 14, 310 becomes 103, and 80 becomes 08), and the meaning would not be lost. However, in the case of digrams, the opposite effect is observed. This already suggests that Voynichese is a cipher based on Roman numerals, as they have the same property: if you take the number 41 = XLI and swap the IXL, you will not get a whole number, as it does not follow the rules of Roman numerals.
This "position effect" manifests itself both at the level of individual semantic units (as in the example with bigrams) and at the level of multiple words and the entire text (this manifests itself in the form of the "prefix-midfix-suffix" pattern). Given the historical context (specifically, the realities of the 15th century, the author's tools, and capabilities), I assume that the Voynich manuscript's cipher is a kind of nomenclature that is additionally encrypted using the letters of an artificially created alphabet (this is not entirely unusual, considering that it was a standard substitution technique for the time, similar to the Theban alphabet), which encrypts both letters and abbreviations.
Why letters and abbreviations?
I came to this conclusion because the list of "letters" I provided is quite flexible. For example, it is easy to see that ch and sh are equivalent (in the sense that words using these letters are almost identical to each other: chey - shey, cheody - sheody, char - shar, chdar - shdar, chckhy - shckhy).
It's not suitable for letters alone, or for abbreviations alone, but it's fine for both.
**
With bigrams, we can conduct a small experiment that shows features that are not typical for substitution. Let's exclude ee(es), ch/sh, and qo from the list. From the remaining list, we can select al-ar and ol-or. We can generate a pair of words by taking a pair like ol-*-in and replacing the asterisk with ai: olaiin, oraiin, and alair. These patterns are not typical for regular substitution, but they align well with the typical numerical cipher-nomenclator. Thus, we should read the words of the manuscript not as words, but as an ordered sequence of numbers, e.g. chey as 50-10-5 (this is not a translation, this just an example of how it might look).
It's inconvenient/impractical/difficult to do!
Maybe that's true, but it's possible, even for the 15th century, plus, given that the manuscript is a product of collective labor, the argument about its bulkiness loses weight, because a group of skilled people would definitely be able to encrypt the text faster than a single person.
It's inconvenient/impractical/difficult to read!
The manuscript is essentially not a work of fiction, but a reference book (all herbals are reference books), which means that the reader does not have to spend time deciphering the entire book. By knowing the key, they can apply it to the relevant pages without any problems.
Apparently, the book was not written for a wide audience (as evidenced by its design and lack of decryption), but for someone who at least knew the key. This suggests that there should have been no difficulties... if the book had fallen into the right hands 
.
...And super-highly interested fact...
Let's look at the top right corner of the You are not allowed to view links. Register or Login to view. page. In the context of the Roman numeral version, it takes on a new meaning...
Rotate the red symbols 90 degrees to the left, and you will get the Roman numeral 102. Such a bold coincidence, and on the first page... Could this be the key?
P.S. I am not an expert, and what I have written may be complete nonsense. The purpose of this post is not to prove that I am right, but rather to reach out to you, to find out your opinion, and to discuss this version together.
I hope this will help whoever deciphers the Voynich manuscript.
Regarding the news from me, there have been no significant advancements in decryption. However, I believe the following is important for progress:
1). Determining the boundary between words in a single line. As we know, Voynichese is a simple nomenclature, and the words (which have been converted to numbers for convenience, but we will refer to them as words) are distorted using a specific alphabet. However, I do not believe that only one word is encoded in a single line, and I do not believe that each word in Voynichese has a decryption on its own. As it appears, after encryption, the word was also "dismembered" into several pieces with the addition of extra elements and may have undergone other manipulations (the position of the numbers was changed so that they were arranged in ascending order, etc.).
How can this be done? To begin with, let's assume that there are a certain number of whole words in one line. Next, we look at the total number of "syllables". Having received their number, we subtract the repetitions (but not all! For example, in a series of chol char chor, we can subtract a maximum of two repetitions, then we get 4 syllables; in the case of daiin daiin, we count only one word daiin, consisting of three syllables - d-ai-in. We don't need a replay).
As a result, we will get the estimated number of syllables, and from there, we can try to calculate the estimated number of words. This is a very imprecise method, and it is unlikely to provide us with any specific data in this form, but we can draw conclusions based on the patterns.
A simpler and more accurate method is to look for consistent word combinations. For example, as I mentioned earlier and as verified by another user, there are many combinations of the form ol.daiin in the botanical section. There are also many combinations of the form chol daiin, but there is also the word choldaiin. We have reason to believe that choldaiin (5 syllables) is a single word.
2). Determine the size of the numbers. We know that certain letters are always in a certain position, as combinations of the form -iin are almost always at the end. By focusing on Voynich A, which is generally easier to work with, we can try to derive approximate ratios of "letters" and "syllables." For example, I can say that ch is larger than e, because e is always followed by ch.
These tasks are not focused on precise results, but rather on identifying any patterns in the system of Voynichese and... just for the sake of research. 
Well, it's time to take a little stock of the work I've done:
1). As for the boundaries of words, it is quite difficult to identify them, and there is no method that is as accurate as possible.
But! We can hypothetically find one of the boundaries and identify the alleged nulls. I consider the boundaries between words to be the words that go beyond the positional rules of Voynichese. For example, there is a reason to believe that
cheody is part of a single word, but
odchy is where one word ends (represented by
od) and another begins (represented by
chy). This makes sense because if the author had permuted all the numerical combinations they obtained after substitution, the result would have been a jumble of confusing anagrams. It would be clearer and easier to mark the boundary between words somehow. In another, words that begin with a typical "ending" (such as
odchy) may be places where the author intentionally did not change the sequence of numbers to indicate the division between words. This means that in such sequences, we will see unchanged pieces of the original words.
As I see it, nulls are highlighted by repetitions. For example, as I have already given an example, in the sequence cha chor chol, the garbage element is most likely ch. It is also possible that the author added entire garbage words, such as
kchorchor, which consist of repetitions and are unlikely to produce any meaningful translation (but this is not a certainty).
That's all I can say here. I can't estimate the average number of words per line and... I haven't made much progress overall. 
2). The scales are a bit more interesting. First, according to positional analysis, the following symbols can be assigned "weights":
o = 1,
ch = 2,
e = 3,
a = 4;
d, l, r = 5;
in, iin, y, m = 6
The weights are assigned according to their position in the words. iin is always at the end, so it has the lowest weight, and o is most often at the beginning, so it has the highest weight. Although I have assigned a common weight to
d,
l, and
r, this does not mean that they are interchangeable or equal to each other.
d occurs 12,000 times,
r - 7000,
l - 9000, plus
r and
can stand after units (
-ir, -il, -iir), and
d does not show such a property.
Briefly, it is easier to assign them one weight based on common properties than to go into these nuances.
So, if these were numbers, we could say that ch is greater than e but less than o; d is greater than a but less than o;
You didn't see all the Voynichese symbols in this list? Yes, I couldn't "count" all the characters, or it would have been pointless. The letter
q and the combination
qo automatically have a weight of 1. The gibbets are mobile, but if we don't consider ligatures like
cTh, I can assume that the gibbets are larger than ch but smaller than
o. The s has different data - it's likely smaller than
e, smaller than
o, and larger than
ch and
a. There are many nuances that prevent us from assigning a weight to this letter (for example,
s often appears after
o but can also appear before it. In general,
s behaves sufficiently mobile to give it a certain weight). Conventionally, it can be equated to 5, but this will be inaccurate.
Total: I don't think I've done anything really worthwhile. I haven't achieved any precise results, but I have made some guesses.
![[Image: PA2IROk.jpeg]](https://i.imgur.com/PA2IROk.jpeg)