- Options
- Login
- Desktop Style
Options
Why and how the text could be Bavarian
- Why and how the text could be Bavarian
-
RE: Why and how the text could be Bavarian
JoJo_Jost > 12-09-2026, 12:57 PM
1. No, I'm not sure if that's it—but many of the patterns in the VMS text make more sense when read from right to left.
2. According to the "graphologist", it was written from left to right on the vellum. But it could have been encrypted from right to left, in which case it would have to be decrypted from right to left.
3. However, the direction in which it was written makes a significant difference, especially in VBM. And so far, translations work a bit better and the generators achieve better scores. But of course, that’s not proof yet. As I wrote, I’m working with both variants. -
RE: Why and how the text could be Bavarian
JoJo_Jost > 14-09-2026, 04:39 PM
We know that in some ciphers, dots and other symbols have served as counters for a single glyph in different versions, even though most of these are more recent. During my attempts at decipherment, I’ve repeatedly noticed the VMS’s high-order counting system.
Here’s one possible version—I don’t know if this has been done before like this:
Here’s how I picture it:
The Gallow K = 1 Loop / looks like a 1 turned to the left = 1
Gallow T = 2 Loops looks vaguely like a horizontal 2
Gallow P = 3 Loops with a lot of goodwill, it looks like a 3.
Gallow f = should have 4 loops, but instead it looks like a 4 turned to the left.
In short: we count up to 4. At similar positions in the slots, the following also appear:
ch = 5
sh = ch + macron = 6
ckh = 7 (sh + 1 Gallow),
cth = 8 (sh + 2 = 8)
cph = 9 (sh + 3) and
sfh = 10 (sh + 4)
Then the families:
the E families:
e=1
ee=2
eee = 3 (then continuing in Roman numerals, o = 6)
eeo =4
eo = 5
o = 6
oe = 7
oee = 8
oeee = 9
oo = 10
The aiin family
an = 1
ain = 2
aiin = 3
aiiin = 4
The oiin family
on = 1
oin = 2
oiin = 3
oiiin = 4
The “ar” family
ar = 1
air = 2
aiir = 3
aiiir = 4
It is interesting to note that the e family often ends in d, while the ai family often begins with d (!) (the oiin family is much smaller, but there exist the d in front, too).
Then there are the ar, al, or, and ol families; here, the counting seems to work through repetition: olol, arar
If we now leave out the vowel bridges (including “qo” and initial “o”), very few glyphs actually remain, along with scattered individual “i”s and others. Actually, mainly just the “s.” (I currently view “m” as an abbreviation for “aiin”).
Theory:
It’s very easy to imagine that these forms are nothing more than table entries. The Gallows and the ch/sh group are the rows, and the families are the columns. Of course, the numbers are in detail arbitrary.
Broadly speaking, if you add the 0th row—that is, without the Gallows and ch/sh groups—and interpret “d” as a switch for two different tables, this would result in about 500 to 900 fields, though not all of them are filled.
This is certainly one possible structure behind VMS.
Yes, I am aware that the identification of the Gallows and the families is, in part, highly interpretive. But fundamentally, the forms exist; the order in which they appear is an open question.
This is just a rough overview for now; in detail, it looks a bit more complex. -
RE: Why and how the text could be Bavarian
Grove > 14-09-2026, 08:17 PM
I’ve wondered if the paragraph initial gallows were part of setting up the cipher for the page. In my mind, the gallows could count 1-4 to inform which table to begin with counting legs and loops… so f=1, k=2, p=3 and t=4.
Then the first character of the word provides the row, etc.., of course trying to establish stage structure of table, row and column proves extremely difficult because certain characters show up in various parts of a word while others are limited to pre, mid and suffix positions. -
RE: Why and how the text could be Bavarian
JoJo_Jost > 15-09-2026, 05:47 AM
@ Grove
According to the LAAFU analysis, the first character of a line does not appear to be part of the first token. I also refer to it as the Line Start Marker (LSM), and the last one seems to have a life of its own as well—the LEM (Line End Marker), though I’m not sure if “m” is simply an abbreviation here.
I’ve examined the LSM closely and so far haven’t found any strong evidence of line encryption. There are some effects in the first line of a paragraph, but I haven’t been able to find a consistent effect across the entire paragraph (which doesn’t mean it doesn’t exist
)
Many of the tokens in the VMS have a clear slot structure. If you really define the first and last letters (qo = 1 atom) as a vowel bridge, what remains at the core is Gallows / BankGallows / ch - sh / - E-family - d.
Similar structures exist in the other families.
However, sometimes other glyphs appear before the Gallows and the Banks, and then there are tokens in which the structure completely shifts.
It does seem, though, that there is a clear basic structure that can explain a lot of the tokens, along with many oddities. -
RE: Why and how the text could be Bavarian
JoJo_Jost > 15-09-2026, 08:03 AM
Here’s the table again, showing just how simple it becomes in REVA (reverse eva)
These are the families that are mapped via the space—that is, from the end of the previous token into this token. It appears that a decision is made here regarding which table (family) to map to. This is then processed (e.g., the e family), followed by additional glyphs until the next vowel bridge.
This is a very clean system. It’s striking that n/l/r are assigned almost equally.
-
RE: Why and how the text could be Bavarian
JoJo_Jost > 15-09-2026, 08:14 AM
Let’s move on to the more complex point. An important criterion of the e-family is that approximately 70 percent of the e-family ends in y (roughly, depending on what you include).
So let’s take a look at the e-family whose tokens end in y!
Here’s the table (click on it, because otherwise its width destroy this post )
We have the 10 glyphs preceding the “e” family that I listed above. Then the main 10 “e” values, and here are the other family values that still exist.
I suspect that since the other clusters are mostly empty and only those in the first two Gallows rows are grouped: Here, the cipherer inserted individual values that he would need later, because his table simply wasn’t consistent (which is no surprise with consonant clusters).
Why did I choose 10 columns? Because it makes sense in other contexts.
Basically, though, the first 6 columns are the E-family clusters that carry the bulk of the data.
Here’s an image of all the values I measured—you can click on it:
The lower rows also list the letters preceding the first 10 glyphs—that is, bigrams and trigrams. Some are grouped into families, among other things.
Here you can see that only the first 6 columns actually contain data. But there are still outliers all the way up to the 9th and 10th columns.
This also shows that the first few rows at the top are indeed anomalies—which makes the theory that they were added later somewhat more plausible. -
RE: Why and how the text could be Bavarian
JoJo_Jost > 5 hours ago
Another indication of a backward-facing structure
I took another closer look at the y-final tokens to understand the logic behind them. I’m still in REVA mode (EVA in reverse). Basis: eva_z3b, all sections without f57v/f116v, only lines with more than two tokens.
I took all tokens ending in -y with at least three glyphs (I count ckh, cth, cph, cfh, ch, sh, and qo as one glyph each), removed the first and last glyphs, then read the maximum run of e/o at the end of the core, preceded by a d if necessary. This yields 9,191 kerns, 4,022 without a d and 5,169 with a d.
Read from the back, the sequence is always the same: y – d (or not) – o (or not) – e-strokes – base glyph.
The ratio of o to d is interesting. Runs containing an o (usually at the end) have a d in 1,047 out of 1,146 cases—that is, 91 percent. Sequences consisting solely of e’s have a d in 51 percent of cases. The o thus strongly attracts the d, almost doubling its frequency. There is no rule without exception in the VMS; 99 o-sequences appear without a d, but the difference is clear. And the two d-branches are not writing variants of each other: the basic glyphs beneath them are distributed differently.
What strikes me about the diagram is the branching. Read backward, it grows step by step: y is one of a few keys, d is a decision with two outcomes, o/e again two, the e-number three, the basic glyph ten. From the general to the specific. That is the order in which one queries a nested table: for example, first the table, then the half, then the column, then the row. Read forward, the most specific element comes first and the key last.
While this doesn’t prove a writing direction, it does show that the more logical order of the characters runs from the end of the token to the beginning of the token—at least when assuming a structure. One might object that “od” is simply a unit, in which case it would also be unremarkable when read forward. But the issue extends beyond the token boundary:
After “ed y,” the next token begins with “qo” in 38 percent of cases; after “od y,” only in 25 percent, just as after “e y” without “d.” Whether there is an “o” in the “e” sequence therefore depends on how the next word begins.
Read forward, this means: A marker inside a token changes how the next token begins. This needs to be explained. Read backward, it’s simpler: in that case, “qo” precedes the “y” token, and the rule becomes: “qo” is most often followed by “ed,” and less often by “od.” A choice based on what was just written—just as any cipher works.
I know that, statistically speaking, this kind of dependency between neighbors can be described in both directions. The difference lies in the simplicity: In one direction, “depends on the predecessor” is sufficient; in the other, you need “depends on the successor.” A writer knows the predecessor—to know the successor, he must know what comes next.
(And this is now the second strange long-range effect in the E family (a “ch” or ‘sh’ before the E family also affects the number of “qo” in the next token).