From “Nymph Baths” to a Cycle Calendar: A Unified Process Hypothesis for the Voynich Manuscript
Posted: Thu Feb 05, 2026 10:06 am
1) What we have built so far (high-level)
We developed a working translation framework for parts of the Voynich Manuscript using two complementary methods:
Bridge Test (Jaccard on UNKNOWN/OTHER vocabulary)
We compare consecutive folios by looking only at “unknown” tokens.
A very low Jaccard continuity indicates a hard register/section change (a new vocabulary block / different mode of writing).
Rubik-Cube Test (Anchor/Faces + neighborhood docking)
We tag certain recurring token families as anchors (TIME, CHECK, SEAL, UNION, FAIL, REST, and two “media” groups).
For each unknown token, we measure which anchors it appears near (window ±6).
Tokens that strongly “dock” to one anchor face behave like operators/parameters, not normal nouns.
This produced a surprisingly consistent operational grammar: the text is structured like a procedure language, with explicit markers for time windows, checks, sealing, joining, resting, and failure/risk.
2) Our working “grammar operators” (anchors)
These are not claimed as final meanings — they are role-stable labels that keep recurring with consistent behavior:
TIME: ote* / qote* family
→ marks time windows / cycle slots (in the stars section) and waiting windows (in the baths/procedure section)
CHECK: chck* / okchy / check* family
→ “inspect / verify / see if it is right”
(we preferred medieval-tone English: “see to it / inspect / examine”)
SEAL: chedy / shedy / chdy family
→ “seal / fix / close / make firm”
UNION: *tam / otam / qotam family
→ “join / graft / unite / set together”
REST: sar* family
→ “let rest / pause / allow to settle”
FAIL / RISK: chor / khos / otchor family
→ “spoilage / wrong state / do-not / it turns”
A (green medium): otal / otol group
→ we treated as a green nutrient medium (in our working narrative: “goat-milk-like nutritive medium / extract”)
Important: we kept it as a green medium concept, acknowledging it could be something broader than literal goat milk.
B (blue water / rinse path): keey / okey / okedy / okeey / qokey... group
→ “water / rinse / flow solution” (blue)
And additional stable object clusters in our plant/hydro model:
dar → “root/base piece”
shol/sheol → “stem segment”
okal/qokal... → “basin/tray/tank”
cheol/chol/dcheol... → “passage / outlet / channel / drain”
3) The key discovery: Two “registers” (writing modes)
Register A — Full procedural mode
Contains many anchors: TIME + SEAL + CHECK + UNION + REST + FAIL + media references (A/B).
This reads like full “do X, seal, check, wait, proceed” instructions.
Register B — Compressed marker mode
Some folios drop SEAL/CHECK/UNION almost entirely and become mostly TIME + B + FAIL + REST.
This behaves like notes, schedules, warnings, or compressed protocol rather than full instructions.
A major proof point was the hard bridge break (example: f86v6 → f87r, extremely low Jaccard continuity), indicating a strong shift in local vocabulary and marker distribution — consistent with a mode change.
4) The process we reconstructed (as a single coherent workflow)
Below is the full operational process we arrived at, integrating plants + media + time windows + hydro cascade logic:
Step 1 — Prepare plant parts (root + stem)
Cut/prepare root/base piece (dar) and stem segment (shol/sheol).
Keep inner surfaces and channels clean (“inside” words: aiin/daiin repeatedly appear after time markers).
Step 2 — Soak in green medium (A)
Place the cut interface into green medium (otal/otol concept).
Do not exceed the time window (TIME markers govern this).
Step 3 — Rinse with blue water (B)
Remove and rinse/spool with blue water (keey/okey-family).
Clean passages/outlets (cheol/chol) and check for spoilage.
Step 4 — Inspect (“CHECK”)
At “inspection time” (TIME + CHECK adjacency), examine whether the state is correct.
If failure markers appear (chor/khos), treat as risk/spoilage.
Step 5 — Join (graft) and seal
Unite the prepared parts (UNION: *tam).
Fix/close/seal (SEAL: chedy/shedy).
Let it settle (REST: sar).
Step 6 — Place into a multi-tier hydro system (cascade)
We proposed a practical mechanical interpretation consistent with repeated “basin / channel / drain / water” language:
A top tray/basin receives blue water (and occasionally green medium in small dose).
Water flows downward tier by tier through passages/outlets.
It collects in a bottom basin, from which it can be:
swapped upward (circulated), or
drained and reset (fresh set-up).
Step 7 — Schedule by cycle slots (stars section)
The star/zodiac pages supply when to do checks, rinses, and risk handling — not only “how.”
5) The Star/Zodiac section (Sternenteil): what it does
From our Rubik analysis, the star section behaves like a cyclical calendar / protocol clock:
ote* tokens appear as cycle ticks / slot labels, not just “minutes.”
There are “slot names” and “slot transitions,” reinforced by ring layouts and repeated segmentation.
Slot types we extracted
We classified time windows into three operational types:
NORMAL slots
routine flow and maintenance
CHECK slots (TIME close to CHECK tokens)
“inspection time / time to examine”
RISK slots (TIME close to FAIL tokens or with embedded FAIL, e.g., oteochor)
“risk time” where spoilage is more likely; triggers stronger rinse/drain logic.
This is why the star section is so powerful: it gives you a schedule, including risk windows.
6) The Bath section (Badeteil): what it does
In our working model, the bath section is not literal bathing for humans — it reads like a friendly, illustrative explanation of internal flows and “healing” behavior, but functionally maps onto a procedure manual:
It uses the same operator vocabulary: TIME → rinse → inspect → seal → rest → (risk handling).
The imagery of “creatures” and colored liquids can be treated as a didactic metaphor:
blue = water/flow/rinse path
green = nutritive medium / inner healing medium
The “creatures react and dispose quickly” matches the FAIL/RISK routine:
detect problem → flush → divert/drain to bottom basin → isolate bad part → reset
So the bath section acts like the narrated/mechanistic layer: it explains what happens inside the channels and basins while the procedural tokens encode the operations.
7) How both parts fit into the manuscript as a whole
Here’s the clean relationship we established:
Stars/Zodiac section = WHEN
A cyclical plan: time slots, inspection windows, and risk windows.
Bath section = HOW (and WHY)
The operational mechanics: soaking, rinsing, joining, sealing, resting, flushing, draining.
Plant/“pharma” pages = WHAT
The objects and targets: plant parts (root/base, stem, channels), preparations, and outcomes.
So together, these form a complete “system manual” structure:
WHAT (plant pieces) → HOW (bath/procedure logic) → WHEN (star/cycle calendar)
And the Bridge Test results support this: hard vocabulary shifts correspond to switching from one functional layer (register) to another.
8) Where we are confident vs. where it’s still “working theory”
Strong (role-stable):
TIME / CHECK / SEAL / UNION / REST / FAIL behave like real operators.
Stars section is schedule-like and slot-based.
Some tokens consistently act like “basin / channel / root/base / stem segment.”
Still working assumptions:
“Green medium = goat milk/extract” is our narrative placeholder.
What’s strong is the two-medium system (green vs blue) and its procedural role, not the literal substance.
Conclusion:
Our analysis suggests the Voynich text behaves less like free prose and more like a structured procedural language. Using a Bridge Test (unknown-vocabulary continuity) we can detect sharp “register switches,” and with a Rubik-style anchor test we can show that recurring token families act like operators (TIME windows, inspection steps, sealing/fixing steps, joining/grafting steps, resting phases, and failure/spoilage flags).
Within this framework, the Zodiac/Star section functions as the timing layer (“WHEN”): it organizes actions into cyclical time-slots, including inspection slots and risk/spoilage slots. The Bath section functions as the mechanics layer (“HOW”): it encodes rinsing/flow control, joining and sealing, resting, and emergency flushing/diverting when spoilage markers appear—often in a friendly, illustrative style that can be read as a metaphor for internal plant-fluid processes. The plant/pharma material provides the object layer (“WHAT”): stem/base pieces, channels, basins, and fluids.
This is not a final decipherment, but it’s a workable, internally consistent process model that makes testable predictions: time-words should recur as slot labels in the star pages; check/fail markers should cluster around specific slots; and procedure pages should mirror those slots with rinse/inspect/seal/rest routines.
Thanks for reading an pin your thoughts into the comment section,
Bastian
We developed a working translation framework for parts of the Voynich Manuscript using two complementary methods:
Bridge Test (Jaccard on UNKNOWN/OTHER vocabulary)
We compare consecutive folios by looking only at “unknown” tokens.
A very low Jaccard continuity indicates a hard register/section change (a new vocabulary block / different mode of writing).
Rubik-Cube Test (Anchor/Faces + neighborhood docking)
We tag certain recurring token families as anchors (TIME, CHECK, SEAL, UNION, FAIL, REST, and two “media” groups).
For each unknown token, we measure which anchors it appears near (window ±6).
Tokens that strongly “dock” to one anchor face behave like operators/parameters, not normal nouns.
This produced a surprisingly consistent operational grammar: the text is structured like a procedure language, with explicit markers for time windows, checks, sealing, joining, resting, and failure/risk.
2) Our working “grammar operators” (anchors)
These are not claimed as final meanings — they are role-stable labels that keep recurring with consistent behavior:
TIME: ote* / qote* family
→ marks time windows / cycle slots (in the stars section) and waiting windows (in the baths/procedure section)
CHECK: chck* / okchy / check* family
→ “inspect / verify / see if it is right”
(we preferred medieval-tone English: “see to it / inspect / examine”)
SEAL: chedy / shedy / chdy family
→ “seal / fix / close / make firm”
UNION: *tam / otam / qotam family
→ “join / graft / unite / set together”
REST: sar* family
→ “let rest / pause / allow to settle”
FAIL / RISK: chor / khos / otchor family
→ “spoilage / wrong state / do-not / it turns”
A (green medium): otal / otol group
→ we treated as a green nutrient medium (in our working narrative: “goat-milk-like nutritive medium / extract”)
Important: we kept it as a green medium concept, acknowledging it could be something broader than literal goat milk.
B (blue water / rinse path): keey / okey / okedy / okeey / qokey... group
→ “water / rinse / flow solution” (blue)
And additional stable object clusters in our plant/hydro model:
dar → “root/base piece”
shol/sheol → “stem segment”
okal/qokal... → “basin/tray/tank”
cheol/chol/dcheol... → “passage / outlet / channel / drain”
3) The key discovery: Two “registers” (writing modes)
Register A — Full procedural mode
Contains many anchors: TIME + SEAL + CHECK + UNION + REST + FAIL + media references (A/B).
This reads like full “do X, seal, check, wait, proceed” instructions.
Register B — Compressed marker mode
Some folios drop SEAL/CHECK/UNION almost entirely and become mostly TIME + B + FAIL + REST.
This behaves like notes, schedules, warnings, or compressed protocol rather than full instructions.
A major proof point was the hard bridge break (example: f86v6 → f87r, extremely low Jaccard continuity), indicating a strong shift in local vocabulary and marker distribution — consistent with a mode change.
4) The process we reconstructed (as a single coherent workflow)
Below is the full operational process we arrived at, integrating plants + media + time windows + hydro cascade logic:
Step 1 — Prepare plant parts (root + stem)
Cut/prepare root/base piece (dar) and stem segment (shol/sheol).
Keep inner surfaces and channels clean (“inside” words: aiin/daiin repeatedly appear after time markers).
Step 2 — Soak in green medium (A)
Place the cut interface into green medium (otal/otol concept).
Do not exceed the time window (TIME markers govern this).
Step 3 — Rinse with blue water (B)
Remove and rinse/spool with blue water (keey/okey-family).
Clean passages/outlets (cheol/chol) and check for spoilage.
Step 4 — Inspect (“CHECK”)
At “inspection time” (TIME + CHECK adjacency), examine whether the state is correct.
If failure markers appear (chor/khos), treat as risk/spoilage.
Step 5 — Join (graft) and seal
Unite the prepared parts (UNION: *tam).
Fix/close/seal (SEAL: chedy/shedy).
Let it settle (REST: sar).
Step 6 — Place into a multi-tier hydro system (cascade)
We proposed a practical mechanical interpretation consistent with repeated “basin / channel / drain / water” language:
A top tray/basin receives blue water (and occasionally green medium in small dose).
Water flows downward tier by tier through passages/outlets.
It collects in a bottom basin, from which it can be:
swapped upward (circulated), or
drained and reset (fresh set-up).
Step 7 — Schedule by cycle slots (stars section)
The star/zodiac pages supply when to do checks, rinses, and risk handling — not only “how.”
5) The Star/Zodiac section (Sternenteil): what it does
From our Rubik analysis, the star section behaves like a cyclical calendar / protocol clock:
ote* tokens appear as cycle ticks / slot labels, not just “minutes.”
There are “slot names” and “slot transitions,” reinforced by ring layouts and repeated segmentation.
Slot types we extracted
We classified time windows into three operational types:
NORMAL slots
routine flow and maintenance
CHECK slots (TIME close to CHECK tokens)
“inspection time / time to examine”
RISK slots (TIME close to FAIL tokens or with embedded FAIL, e.g., oteochor)
“risk time” where spoilage is more likely; triggers stronger rinse/drain logic.
This is why the star section is so powerful: it gives you a schedule, including risk windows.
6) The Bath section (Badeteil): what it does
In our working model, the bath section is not literal bathing for humans — it reads like a friendly, illustrative explanation of internal flows and “healing” behavior, but functionally maps onto a procedure manual:
It uses the same operator vocabulary: TIME → rinse → inspect → seal → rest → (risk handling).
The imagery of “creatures” and colored liquids can be treated as a didactic metaphor:
blue = water/flow/rinse path
green = nutritive medium / inner healing medium
The “creatures react and dispose quickly” matches the FAIL/RISK routine:
detect problem → flush → divert/drain to bottom basin → isolate bad part → reset
So the bath section acts like the narrated/mechanistic layer: it explains what happens inside the channels and basins while the procedural tokens encode the operations.
7) How both parts fit into the manuscript as a whole
Here’s the clean relationship we established:
Stars/Zodiac section = WHEN
A cyclical plan: time slots, inspection windows, and risk windows.
Bath section = HOW (and WHY)
The operational mechanics: soaking, rinsing, joining, sealing, resting, flushing, draining.
Plant/“pharma” pages = WHAT
The objects and targets: plant parts (root/base, stem, channels), preparations, and outcomes.
So together, these form a complete “system manual” structure:
WHAT (plant pieces) → HOW (bath/procedure logic) → WHEN (star/cycle calendar)
And the Bridge Test results support this: hard vocabulary shifts correspond to switching from one functional layer (register) to another.
8) Where we are confident vs. where it’s still “working theory”
Strong (role-stable):
TIME / CHECK / SEAL / UNION / REST / FAIL behave like real operators.
Stars section is schedule-like and slot-based.
Some tokens consistently act like “basin / channel / root/base / stem segment.”
Still working assumptions:
“Green medium = goat milk/extract” is our narrative placeholder.
What’s strong is the two-medium system (green vs blue) and its procedural role, not the literal substance.
Conclusion:
Our analysis suggests the Voynich text behaves less like free prose and more like a structured procedural language. Using a Bridge Test (unknown-vocabulary continuity) we can detect sharp “register switches,” and with a Rubik-style anchor test we can show that recurring token families act like operators (TIME windows, inspection steps, sealing/fixing steps, joining/grafting steps, resting phases, and failure/spoilage flags).
Within this framework, the Zodiac/Star section functions as the timing layer (“WHEN”): it organizes actions into cyclical time-slots, including inspection slots and risk/spoilage slots. The Bath section functions as the mechanics layer (“HOW”): it encodes rinsing/flow control, joining and sealing, resting, and emergency flushing/diverting when spoilage markers appear—often in a friendly, illustrative style that can be read as a metaphor for internal plant-fluid processes. The plant/pharma material provides the object layer (“WHAT”): stem/base pieces, channels, basins, and fluids.
This is not a final decipherment, but it’s a workable, internally consistent process model that makes testable predictions: time-words should recur as slot labels in the star pages; check/fail markers should cluster around specific slots; and procedure pages should mirror those slots with rinse/inspect/seal/rest routines.
Thanks for reading an pin your thoughts into the comment section,
Bastian
