The Voynich Ninja

Hi, my name is Sean aka TheVisad, and over the course of the last several months I have been working on the manuscript. Instead of approaching the systems as a language, I decided to look at it from a perspective that it was a data system instead. I have long worked with data systems and saw this as just something fun to poke at. Outlined below, is what I have discovered so far regarding the Manuscript. 

By no means is this a decipherment, from the moment I picked up the book I knew it was not translatable. This is a computational demonstration that the Voynich Manuscript is a deterministic, reproducible structural system. That every token of it's 37,671 word corpus can be derived from first principles with zero residual entropy. That the machine is based in the Ottoman Anatolian Turkish Eastern Arabic abjad system and used as mathematical formula based on mod35, which has historical aspects in the Eastern systems. That the machine is a two-sided wheel that provides the triple-layer word structure that maps directly to the only period in Turkish linguistic history where the language simultaneously used all three morphological systems.

What I propose is that the Voynich manuscript is the operational log of a 14 selector volvelle. The Voynich Manuscript is the output of a medieval Islamic pharmaceutical look up device used by the letterists in the Ottoman Empire in the 1420's. Voynich is not a natural language, it is not a cipher over a hidden plaintext, and it is not a meaningless hoax. The manuscript itself tells us the story in several of the folios. You are not allowed to view links. Register or Login to view. IS the machine described below, the 9-page fold out IS THE internal mechanism of the volvelle. The four gated dial on You are not allowed to view links. Register or Login to view. IS THE gallows. The pages outline the machine itself, and it's not until we the "Frankensteinian Morphology" of the three linguistic families, Persian Prefixes, Arabic Interiors and Turkish Agglutinative Suffixes that the system clicks. 

The device consists of 14 mechanical selectors, arranged into two functional layers, the machine, and apparatus layers. The machine layer is a 14-selector deterministic finite state machine with four outer wheels, eight interior position dials, a walk state selector derived from sequential context, and a binary tashdid doubling operator. This machine layer generates all tokens, with zero residual entropy at 100% token determination. The interior dials operate via a linear congruential generator over residues modulo 35, where the gallows detent selects the multiplier. Four specific abjad numerical positions (q=Qaf=19, r=Ra=20, s=Sin=15, y=Ya=10) match the Eastern Arabic abjad system exactly, confirming the arithmetic base.

The apparatus layer is a 13-slot cyclic ring machine operating concurrently with the machine layer. The apparatus labels correspond to ring positions clockwise from an output gate: collection, preamble, reset gate, cosmological transformation mediator, transformation operator, herbal routing, cosmological routing, cycle-completion gate, and five vapor/formula tracking labels.

The machine's structure maps to the medieval Arabic za'irajah, a mechanical divination device documented by Ibn Khaldūn in 1377. The za'irajah has 8 concentric circles, 12 zodiac spokes producing 13 operational states (12 zodiac chords + 1 pole/vers-clef), a 55 × 131 lookup table (7,205 cells), and a fixed operating poem by Mālik b. Wuhayb that indicates the description and operation of the controls  for a full cycle completion. These are matching, respectively, the volvelle's 8 prefix channels, 12 active apparatus slots + 1 K-GATE pole, ~4,243-entry dictionary (interior lookup space ~7,861 cells), and the k-label cycle-completion gate.

I propose that the Voynich manuscript, is the mechanical output, of a paper machine, created by two medieval Islamic scribes. These scribes worked on the manuscript sometime in 1400-1420, and wrote the book based on the output of the machines deterministic view. The manuscript we have is the written record of this process. There are no visible mistakes, there are no visible cross outs, and no observed locations that the scribes ever made a mistake. Across the entire 37,671 tokens, the manuscript presents a clean execution record, not a drafted prose text.

I have additional theories that are associated with the OAT morphological aspects, and how they correlate to medical purposes. I am pursuing a line of research that the pages are an actual recipe. Not in the normal sense that you and I are familiar with, grab two teaspoons of salt to do this with it. Rather, they are a machine record of a process to take a set of ingredients and use the machine to determine the process. The users of the za'irajah created a question; this question was broken down and used as the input of the wheel. Then following the guidelines of the poem, allowed for one full cycle of the device to obtain its output. The output of the device was the medical recipe that divination provided from the device and was used to “cure” the patient. 

The attached images show some of the output from the scripts I have created, and the zip file contains a demo of the wheel's processes as it moves through the folios. I welcome any and all questions, feedback, and criticism. I have a ton of research, files, data, scripts, grammar output, and more. I would welcome working with someone to verify the data, the machine, and the findings I have. I am still assembling the public repository, but have a private repository I can share readily and over 2tb of data offline that can be shared that is directly related to overall project. 

Sean Rice aka TheVisad

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

As indicated in my previous post, the Volvelle that seems to have generated the book tokens is based on or is a variation of the "Za'irajah of the world". Using the Jaccard we can compare the architectural features of the two devices. However, why stop there? As a control, we should include as many medieval computations devices as possible, the astrolab, the apian volvelle, the alberti cipher wheel and others. Permutation tests were done across all 9 different devices. The nearest scores a 0.625, while the Za'irajah scores a Jaccard value of 1.0. The zairaja, which is a variation of the za'irajah scores incredibly high in the matrix as well. This shows that mathematically, the devices share the same features and functionality.




For the healer, there was a procedure to follow, a poem to be sung. This is step one and cannot be stressed enough that this requirement must be met. Ask the question in rhyme, get an answer in rhyme. Otherwise you will receive a "normal" answer. This meant that a true divinational request required that the poem be sung and rhymed. This was required for the mechanical aspects of the wheel. 28 Arabic letters and 7 wheels for celestial bodies equals the mod 35 that the machine is based on. Without this rhyme structure, the arithmetic does not provide a coherent answer. Ibn Khaldun explicitly stated, "Without the bayt comparison, the method does not produce a rhymed answer."

Step two consists of resolving the poem into the fixed carrier that the original pangrammic provides, all 28 letters exactly once. The machine will then cycle through both the question letters and the chord letters and compare them against this verse. The shaddah, or the tashdid is one of the diacritics indicating a consonant doubling, making the total letter count 43, not 28. This is the mechanism that fires in my model, the shadda and the tashdid are the same mark, name and function in the device. The carrier poem is shown below and is in the pdf’s list in the end. This poem is required to assemble the question for the individual.

سؤال عظيم الخلق حزت فصن إذن
غرائب شك ضبطه الجد مثلا

Step three is where the za'irajah adds in the 7 astral or planetary governors, the seven classical planets, the sun, moon, mars, mercury, jupiter, venus and saturn. We accept the original question in the form of 28 and then add in the 7 dials here to receive our modulus 35 for the emission. Mod 35 is the alphabet and the heavens.

In step four these letters become numbers, which is where the abjad comes into play. Each Arabic letter has a fixed numerical value under the Eastern abjad which it appears the Voynich volvelle is based on. qaf = 100, ra = 200, sin = 60, ya = 10 and the 28 letters collapse to 5 class values under mod 35, absorb (0), vowel (10), gallows (15), conson (25), and process (30). These are required for step five where we drive the machine using these values. See the image for the look up table, while hard to see, shows roughly ~7200 internal values.

We are at the stage where we need to drive the machine. Step five allows the LCG to mechanically advance the state of the machine. The gallows detent physically selects the multiplier. With no gallows, a = 20 and we have a 2 cycle vowel <-> consonant. With GtB gallows, a = 6 and has the same 2 cycle process. In GkB gallows, a = 24 and this allows the machine to escape out of the sequence and process into a different orbit. Finally at GkM gallows, a = 3, and we are in our process attractor mode of the machine. This two cycle process is modular arithmetic, and not a vowel consonant alternation. It produces this pattern because 10x20 mod 35 equals 25 and 25x20 mod 35 is 10. This pattern was observed in the manuscripts linked in the post.

The last stage is the divinational aspect of the wheel. This is where our answer appears in the system. The cycle class above allows us to select the value class, and the outer wheels which provide the prefix, suffix and walk state select the specific grapheme within that class. That is the two stage look up, the LCG gives you the row, the outer selectors give you the column. In the za'irajah the cycle arithmetic gives you the index, this is entered into the pole verse and whatever letter appears you write it down. Do this 13 times, with a repeating cycle and you get the output phrase. For the Voynich, we have the same process by the 28 letters expanded to 90 suffix positions and the output is a pharmaceutical recipe line instead of a rhymed divinatory answer. Again, this recipe line is not one we are likely familiar with and is just the mechanical output of the machine. For the Voynich Volvelle we receive the tokens from the Voynich manuscript as an output and which are assembled from the viewing area.

So if we know that this "Za'irajah of the world", is the driving force behind the Voynich volvelle and is providing an answer, then how can we use this? How can we verify that what we are seeing from this machine is telling us to use these methods? Can we verify these values out of the machine correspond to medicines? Compounds? Seasons? We cannot translate it, we do not know what it is saying, so how can we be sure that what we are seeing is really an output of the device? One method I am exploring is  using the old manuscripts from the OAT era that were specifically for medical purposes. Taking the core fact that the machine is built from the tri lateral roots that form the OAT language, it gives a targeted approach to look at the building blocks that might make the tokens. Using the words with unique consonants that match the consonant skeleton that is being drawn from the machine, the Jaccard matches overwhelmingly indicating medical recipes that line up with very specific compounds within the manuscripts. This is a direct correlative aspect from the Za’irajah and can be looked up in the books listed below.

As part of the battery of adversarial tests the Codex Seraphinianus was used to determine the Jaccard closeness to the consonant skeleton.




In the attached images in this chapter are the Za’irajah wheel and the look up table. The look up table has been feature matched to be able to target the letters and content inside of the table for any potential recovery. The Jaccard similarities for the currently replicated Voynich Volvelle and the other devices similar to it and the operating procedures. There is an open scad rendering of the machine itself.



Note: the volvelle viewer in the previous post is an automated viewer for replaying the original folios through the Vonich Volvelle and is based on an earlier version without the apparatus layer. This layer can be seen in the OpenScad model of the machine. The OpenScad model represents the current development path and the viewer a presentable, functional model that explains the Voynich Volvelle.

Thanks for your time reading this post. Please feel free to send questions or comments.


Sean Rice


















Research books used to identify and discover the above information. This data comes from real history and real books that indicate these historical aspects

Prior art for the Za’irajah of the world, its operating procedures and poems showing a historical reference. I have reached out to the library in Morocco which holds the manuscript that has the personal account of the device itself by its creator. There are no digitized versions of this manuscript and this would be the final closing aspect on the Za’irajah. There is first hand information on the device itself from the books we have from Franz Rosenthal. Those links are listed below. 

Requested Manuscript:
Bibliothèque Nationale du Royaume du Maroc / former Bibliothèque générale de Rabat
Manuscript inventory no. 491
Catalogued by Évariste Lévi-Provençal as no. 478 in Les manuscrits arabes de Rabat, Première série, 1921
Section: Astrologie et Sciences occultes
Composite manuscript / recueil
Target item: II. Ouvrage de gadwal by Gamâl/Jamāl ed-Dîn Abd el-Malik b. Abd Allah el-Marjānī
Folios 7–29



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

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

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

Hi Sean,

Thanks for the detailed write-up and for sharing the viewer. I spent some time with the zip file and tried to reproduce the central claim. A few notes, roughly in order of how central they seem to the framework.
On C10: "100% slot purity at lag=-1, 37670/37671 tokens"
I tested this against a ZLZI based transliteration with a four-slot prefix/gallows/core/suffix decomposition (37,465 tokens; within-line adjacency, n = 32,303 pairs). The natural readings I could derive from your description:

gallows[t] == gallows[t-1] (literal identity): 43.3%
gallows[t] given token[t-1] (best class given previous token): 66.4%
gallows[t-1] given (prefix, core, suffix) of token t: 64.9%
has_gallows[t] == has_gallows[t-1] (binary presence): 53.2%
gallows[t] == prefix[t+1] (literal display-register reading): 13.8%
shuffled-gallows null: 65.0% (SD 0.10 pp), so the best reading is 1.4 pp above null
naive first-character baseline given previous token: 33.4% entropy reduction, slightly stronger than the gallows version at 29.9%

The best reading I could construct is about 33 percentage points below 99.997%. The "1 exception out of 37,671" precision is also the part I'd most like to see derived: categorical data on a 9-value column doesn't naturally produce exactly one exception under any test I could construct from the text. If MACHINE_PROOF.md §8 is shareable, the rule and the result would both be straightforward to verify.
On manuscript_data.js

The file contains a type-consistent VMS to English substitution dictionary: 7,861 distinct word types, each mapped to one fixed English gloss, with zero word types showing more than one gloss and only 20 of 7,861 showing role variation. Multiple distinct VMS tokens collapse onto the same gloss: daiin = aiin = "reference"; chedy = shedy = "and"; ar = dar = "seek"; al = dal = "take"; qokeey = qokedy = "transform".
Applied to You are not allowed to view links. Register or Login to view. line 1 the result is "which seek type marker become reservoir formula reading ember become". Applied to You are not allowed to view links. Register or Login to view. line 1 it is "vessel purification ember the opening trace open vessel extract toward pathway vent outward quench fire vessel opening ember sustains". The output is not coherent English at the sentence level.
The forum post says "we cannot translate it"; the viewer file is, functionally, a complete word-by-word translation. Those two framings are contradictory, and it would help to know which one is current.

On the execution graphs
Role assignment in the data file is fixed per word type: every daiin is LBL, every qokeey is PRC, with only 20 of 7,861 types showing any variation. That makes the "FULL_EXECUTION" graphs a token sequence with one colour per word type rather than a per-position computation. If there is machinery that produces context-dependent role assignment, the token-to-role rule would be the thing to share, because the current visualisation cannot show it.

On the za'irajah comparison
Several of the 20 features in the Image 4 matrix ("Pharmaceutical use", "Humoral degree encoding", "Abjad numeral system", "Contemporary 1400–1440 CE") align with the proposed reading of the VMS by construction rather than independently. A version of the test restricted to device-independent geometric and procedural primitives, pre-registered before the VMS comparison, would be more informative.
Two further concerns with the historical comparator. David Link's reconstruction of the as-Sabti zairja (Scrambling T-R-U-T-H: Rotating Letters as a Material Form of Thought, in Variantology 4, eds. Zielinski and Fuerlus, Cologne: König, 2010) treats the procedure as deterministic and rule-governed but produces only a partial reconstruction; he writes in his Section 10 that he was unable to determine a coherent procedure for the reverse-table operation in the later cycles. The single example output in the Muqaddimah (the Idris / Holy Spirit answer) is one rhymed verse of around 47 characters produced from one question, not a 37,000-token continuous corpus, so the determinism claim and the corpus-generation claim are at different scales. Rosenthal's footnote (Muqaddimah trans. Rosenthal vol. 3, p. 197, fn. 880) describes the rationale of the letter-determination procedure and its relationship to the table as "by no means clear to me", and Link's reconstruction inherits that uncertainty for parts of the procedure.

I have also not found a primary-source attestation of the zairja as a pharmaceutical or medical device. Link's account, the standard scholarly reference, treats the device throughout as divinatory. If you are drawing on a medical-zairja tradition, a citation would help locate it.

Minor
The token count differs across your own materials. The forum post and the OpenSCAD comments cite 37,671; manuscript_data.js (generated 2026-04-07) gives 37,046. The C10 denominator does not match the data file. Worth reconciling.
None of this is a claim that the underlying structural intuitions are not worth pursuing. The four-slot decomposition in your wheel_spec (8 prefixes, 4 gallows, 9 named suffixes) tracks with what I've been working on independently - I'm curious how you landed on this configuration. The questions above are about the deterministic-generator claim and the za'irajah linkage specifically. If you can share the §8 rule and the source transliteration, I will run the same tests against your numbers and post the result.
Best,

Hi Ed,
Thank you for reading through my post and going into detail on the presentation and replication issues you are experiencing. You beat me to the next post, and asked the right questions here for the next section.

The “translation” framing in the viewer was an early idea that morphological words could be attached to the tokens once they were decomposed to the interior structure. I realized quickly that this was an incorrect interpretation and removed the visual layer, but left it there programmatically. It should have been stripped completely. The viewer should simply be looked at as an early demo and not as the canonical proof layer. Any content layer work is now kept separate from the machine and apparatus layers.
You are correct that the role assignment is effectively 1 to 1 at the word type level. This is expected, and not necessarily a gap. Each word is a concatenation of the selector state: word_t = concat(prefix, gallows, dial0,dial1,..., dialN, suffix). The word is the machine state readout, and not a lookup into it. The determinism chain shows entropy collapsing from 13 bits to 0 as each selector is added. By the time decomposition is complete, exactly one word can appear at that position and one role. The underlying concentric decomposition is not original to my work, Stolfi’s 2000 analysis identified the same layered word structure (crust-mantle-core) from frequency analysis. What I extended was his resolution, those three layers decompose further into the 8 prefix positions, 90 suffix positions and 8 interior dial positions with LCG arithmetic.

For the C10 Slot Purity test, MACHINE_PROOF.md §8 is core to some of the additional work and theories I have. This will help validate some of the theories I am still making regarding this aspect. Please see the attached MACHINE_PROOF S8.txt regarding this request.  Please let me know if you have any follow up questions or feedback on this additional data that I can clear up.

The token discrepancy is related to my adversarial approach in the corpus and testing methodologies. I maintain a copy of the IVTFF and the Voynichese.com transcriptions and run the battery of tests on both. IVTFF has 37,671 tokens and tested with lag -1 it hits 37,670 / 37,670. Voynichese.com maintains a corpus of 37,886 raw tokens. I filtered out the non Voynich tokens (?, !, etc) from the Voynichese.com version and lines < 2 to obtain the final token count of 37,046 for testing purposes. During lag testing at -1, the normalized Voynichese.com corpus gives 37,045/37,045. This does make it difficult to present a single number appropriately, and I will find a way to correct this in the presentation layer. The Voynichese.com layer represents the majority of the early work, before settling to IVTFF as the primary corpus.




For the za’irajah comparison, I have two matrices, one for the architectural features, which has 16, and one for the procedural and operational procedures from the Muqaddimah which contains 20 features. I referenced one image and then posted the second image. I added the other image here for reference. The references for the abjad encoding, the pharmaceutical use, the input process, the way to decipher the lookup tables, specific look up table navigation all are coming from the Muqaddimah. Without the physical za’irajah to verify I am feature matching off of these values that the book calls out. The modular arithmetic (LCG D_{k+1} = a·D_k mod N), multi tier selectors, the tashdid binary doubler, and elemental letter groups mapped to the detents are the ones that carry the weight. I cite Link’s work on the structural description, but not the full procedure and the incompleteness of his look up table is a fair point. I have been working on a version of the look up table recovery process to some degree of success.





The idea that the za’irajah was a pharmaceutical device or at least had some usage for this aspect and was in scope for the device, came from the Muqaddimah and the operating poem itself. You are not allowed to view links. Register or Login to view. Pages 182 to 187 of the book are the poem itself and page 188 of book three calls out the poem’s usage and the reference mentions the look up table. The title of that is Ibn Khaldun's statement to the operating poem, and called out the science of medicine and the crafting of alchemy. The poem calls out alchemy, the science of medicine, elixirs and mixtures and patients being cured in almost every line and then directly references to the look up table values, making this more probable. Unfortunately, this and the other book I mentioned that is in Morocco are the only true first or even translation references we have that I am aware of. I realize I am just referring back to the same document you point out is a small sample.

The poem for easy reference.

‘O student of medicine and of Jâbir's science (alchemy),⁸⁵⁵
And you who know the extent of the powers in succession,
If you want the science of medicine, there must exist a
    relationship
For the laws of scales, which ⁸⁵⁶ will hit upon a (possible) way.

III, 153

Your patient will be cured. The elixir will be right.
The soundness of the mixtures of your composition will
    be apparent.’


In relation to your independent convergence, I found 9 named suffixes (dy, in, ey, ol, ar, hy, al, ky, am) that came from the original morphological decomposition and one miscellaneous catch-all that handled everything else. The misc accounts for 28.4% of tokens, roughly ~10,506 tokens in the corpus. It contains 81 hidden classes, split by terminal 2 characters of the word. Not of the suffix label, of the full word form, since misc words end at different positions. The disjointness criterion at Jaccard = 0 for all pairs means each 2-character ending maps to a completely distinct interior pool. This confirmed mechanically the distinct wheel positions. Focusing on my theory that this is OAT/Letterist technology in the Eastern abjad, 90 appears firmly in the position of ص (sad) and is present in the za'irajah lookup table headers. In abjad cosmology, both are boundary values but 90 closes tens register and 9 closes the units.

Jaccard disjointness test:
# For each pair of misc sub-classes A, B:
s1 = {interior(w) for w in misc_words_ending_in_A}
s2 = {interior(w) for w in misc_words_ending_in_B}
inter = len(s1 & s2)        # shared interiors
union = len(s1 | s2)        # all interiors combined
J(A, B) = inter / union      # = 0.000 for all 28 tested pairs

The expansion rule is:
def expanded_suffix(word, suffix):
    if suffix != "misc":
        return suffix
    end2 = word[-2:] if len(word) >= 2 else word  # terminal 2 characters of word
    return "misc_" + end2


Thank you,
  Sean Rice

(16-05-2026, 06:07 AM)DG97EEB Wrote: You are not allowed to view links. Register or Login to view.Hi Sean,

Thanks for the detailed write-up and for sharing the viewer. I spent some time with the zip file and tried to reproduce the central claim. A few notes, roughly in order of how central they seem to the framework.
On C10: "100% slot purity at lag=-1, 37670/37671 tokens"
I tested this against a ZLZI based transliteration with a four-slot prefix/gallows/core/suffix decomposition (37,465 tokens; within-line adjacency, n = 32,303 pairs). The natural readings I could derive from your description:

gallows[t] == gallows[t-1] (literal identity): 43.3%
gallows[t] given token[t-1] (best class given previous token): 66.4%
gallows[t-1] given (prefix, core, suffix) of token t: 64.9%
has_gallows[t] == has_gallows[t-1] (binary presence): 53.2%
gallows[t] == prefix[t+1] (literal display-register reading): 13.8%
shuffled-gallows null: 65.0% (SD 0.10 pp), so the best reading is 1.4 pp above null
naive first-character baseline given previous token: 33.4% entropy reduction, slightly stronger than the gallows version at 29.9%

The best reading I could construct is about 33 percentage points below 99.997%. The "1 exception out of 37,671" precision is also the part I'd most like to see derived: categorical data on a 9-value column doesn't naturally produce exactly one exception under any test I could construct from the text. If MACHINE_PROOF.md §8 is shareable, the rule and the result would both be straightforward to verify.
On manuscript_data.js

The file contains a type-consistent VMS to English substitution dictionary: 7,861 distinct word types, each mapped to one fixed English gloss, with zero word types showing more than one gloss and only 20 of 7,861 showing role variation. Multiple distinct VMS tokens collapse onto the same gloss: daiin = aiin = "reference"; chedy = shedy = "and"; ar = dar = "seek"; al = dal = "take"; qokeey = qokedy = "transform".
Applied to You are not allowed to view links. Register or Login to view. line 1 the result is "which seek type marker become reservoir formula reading ember become". Applied to You are not allowed to view links. Register or Login to view. line 1 it is "vessel purification ember the opening trace open vessel extract toward pathway vent outward quench fire vessel opening ember sustains". The output is not coherent English at the sentence level.
The forum post says "we cannot translate it"; the viewer file is, functionally, a complete word-by-word translation. Those two framings are contradictory, and it would help to know which one is current.

On the execution graphs
Role assignment in the data file is fixed per word type: every daiin is LBL, every qokeey is PRC, with only 20 of 7,861 types showing any variation. That makes the "FULL_EXECUTION" graphs a token sequence with one colour per word type rather than a per-position computation. If there is machinery that produces context-dependent role assignment, the token-to-role rule would be the thing to share, because the current visualisation cannot show it.

On the za'irajah comparison
Several of the 20 features in the Image 4 matrix ("Pharmaceutical use", "Humoral degree encoding", "Abjad numeral system", "Contemporary 1400–1440 CE") align with the proposed reading of the VMS by construction rather than independently. A version of the test restricted to device-independent geometric and procedural primitives, pre-registered before the VMS comparison, would be more informative.
Two further concerns with the historical comparator. David Link's reconstruction of the as-Sabti zairja (Scrambling T-R-U-T-H: Rotating Letters as a Material Form of Thought, in Variantology 4, eds. Zielinski and Fuerlus, Cologne: König, 2010) treats the procedure as deterministic and rule-governed but produces only a partial reconstruction; he writes in his Section 10 that he was unable to determine a coherent procedure for the reverse-table operation in the later cycles. The single example output in the Muqaddimah (the Idris / Holy Spirit answer) is one rhymed verse of around 47 characters produced from one question, not a 37,000-token continuous corpus, so the determinism claim and the corpus-generation claim are at different scales. Rosenthal's footnote (Muqaddimah trans. Rosenthal vol. 3, p. 197, fn. 880) describes the rationale of the letter-determination procedure and its relationship to the table as "by no means clear to me", and Link's reconstruction inherits that uncertainty for parts of the procedure.

I have also not found a primary-source attestation of the zairja as a pharmaceutical or medical device. Link's account, the standard scholarly reference, treats the device throughout as divinatory. If you are drawing on a medical-zairja tradition, a citation would help locate it.

Minor
The token count differs across your own materials. The forum post and the OpenSCAD comments cite 37,671; manuscript_data.js (generated 2026-04-07) gives 37,046. The C10 denominator does not match the data file. Worth reconciling.
None of this is a claim that the underlying structural intuitions are not worth pursuing. The four-slot decomposition in your wheel_spec (8 prefixes, 4 gallows, 9 named suffixes) tracks with what I've been working on independently - I'm curious how you landed on this configuration. The questions above are about the deterministic-generator claim and the za'irajah linkage specifically. If you can share the §8 rule and the source transliteration, I will run the same tests against your numbers and post the result.
Best,

Hi Sean,

Amazing.

Where do I start? I don't see an article. No explanations in the viewer. What am I missing?

Hi DG97EEB,

What are you quoting?

Quote:On C10: "100% slot purity at lag=-1, 37670/37671 tokens"
...
The C10 denominator does not match the data file.

What is C10? Am I missing a context to this discussion, like an article on linkedin or somewhere else?

(16-05-2026, 08:49 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.Hi Sean,

Amazing.

Where do I start? I don't see an article. No explanations in the viewer. What am I missing?

Hi DG97EEB,

What are you quoting?

Quote:On C10: "100% slot purity at lag=-1, 37670/37671 tokens"

Am I am missing a context to this discussion, like an article on linkedin or somewhere else?

It's referenced in his zip file

(16-05-2026, 09:07 PM)DG97EEB Wrote: You are not allowed to view links. Register or Login to view.It's referenced in his zip file

Where?

(16-05-2026, 09:28 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.

(16-05-2026, 09:07 PM)DG97EEB Wrote: You are not allowed to view links. Register or Login to view.It's referenced in his zip file

Where?

So this is why I never reply with these studies. LLMs lie... But the funny thing is that he actually does make the claim in the file he just supplied, which is why when he fed my response in to his GPT instance, it recognised it and his reply contained it.. so either Claude has super future powers or it inferred it from other data..

Either way I apologise and will get back to my corner... At this point though, we should really just se you an MCP server and have everyone's agents talk to each other...?

(16-05-2026, 09:43 PM)DG97EEB Wrote: You are not allowed to view links. Register or Login to view.

Quote:

Quote:It's referenced in his zip file

Where?

So this is why I never reply with these studies. LLMs lie... But the funny thing is that he actually does make the claim in the file he just supplied, which is why when he fed my response in to his GPT instance, it recognised it and his reply contained it.. so either Claude has super future powers or it inferred it from other data..
Either way I apologise and will get back to my corner...

It's referenced in the image of the machine in the corner where you referenced the request for the machine proof. You are right about LLMs, they make stuff all the time, it's a core problem in their design. Here is a link to the machine stl, it's printable on a 3d printer.

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


// ============================================================
// assembly.scad  --  Full Volvelle Assembly
//
// UPDATED (2026-05-14): Added Layer B apparatus ring (ring_apparatus.scad).
// The apparatus ring is the outermost layer, 13 slots encoding the
// cyclic execution grammar (Batteries 1-16). Sits above prefix ring.
// Element-taksir label assignments: l=Earth, s=Water, d/k=Fire, a=Air.
// Section profiles: H=s->l (28 deg), C=o->f->x (166 deg), S=l dominant.
//
// TIMING MODEL (2026-04-11):
//  Gallows arm is a COMMITTED-STATE DISPLAY REGISTER.
//  It shows γ_{t-1} (previous-token gallows) while token t is written.
//  Prefix and suffix are synchronous live selectors (read at write time).
//  Full state vector in machine output: O_t = (π_t, σ_t, γ_{t-1})
//  For corpus comparison use raw_gallows[t+1] for corpus position t.
//  Empirically confirmed: 100% slot purity at lag=-1 (37670/37671 tokens).
//  See MACHINE_PROOF.md §8 / Claim C10.

---

(16-05-2026, 09:28 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.

(16-05-2026, 09:07 PM)DG97EEB Wrote: You are not allowed to view links. Register or Login to view.It's referenced in his zip file

Where?

The reference is in the image I shared on the machine openscad in the corner on the notes regarding the timing model.

(16-05-2026, 10:00 PM)thevisad Wrote: You are not allowed to view links. Register or Login to view.

(16-05-2026, 09:43 PM)DG97EEB Wrote: You are not allowed to view links. Register or Login to view.

Quote:

Quote:It's referenced in his zip file

Where?

So this is why I never reply with these studies. LLMs lie... But the funny thing is that he actually does make the claim in the file he just supplied, which is why when he fed my response in to his GPT instance, it recognised it and his reply contained it.. so either Claude has super future powers or it inferred it from other data..
Either way I apologise and will get back to my corner...

It's referenced in the image of the machine in the corner where you referenced the request for the machine proof. You are right about LLMs, they make stuff all the time, it's a core problem in their design. Here is a link to the machine stl, it's printable on a 3d printer.

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


// ============================================================
// assembly.scad  --  Full Volvelle Assembly
//
// UPDATED (2026-05-14): Added Layer B apparatus ring (ring_apparatus.scad).
// The apparatus ring is the outermost layer, 13 slots encoding the
// cyclic execution grammar (Batteries 1-16). Sits above prefix ring.
// Element-taksir label assignments: l=Earth, s=Water, d/k=Fire, a=Air.
// Section profiles: H=s->l (28 deg), C=o->f->x (166 deg), S=l dominant.
//
// TIMING MODEL (2026-04-11):
//  Gallows arm is a COMMITTED-STATE DISPLAY REGISTER.
//  It shows γ_{t-1} (previous-token gallows) while token t is written.
//  Prefix and suffix are synchronous live selectors (read at write time).
//  Full state vector in machine output: O_t = (π_t, σ_t, γ_{t-1})
//  For corpus comparison use raw_gallows[t+1] for corpus position t.
//  Empirically confirmed: 100% slot purity at lag=-1 (37670/37671 tokens).
//  See MACHINE_PROOF.md §8 / Claim C10.

---

(16-05-2026, 09:28 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.

(16-05-2026, 09:07 PM)DG97EEB Wrote: You are not allowed to view links. Register or Login to view.It's referenced in his zip file

Where?

The reference is in the image I shared on the machine openscad in the corner on the notes regarding the timing model.

Excellent..so Claude hallucinated about its own hallucination... I might build that MCP server after all