The Voynich zodiac pages contain one of the manuscript’s most controlled text populations.
Short labels.
One beside each human figure.
Roughly thirty around a sign.
Bounded.
Positional.
Repeated across several rings.
This makes them different from ordinary Voynich prose before we assign a single meaning.
In 2026, Vitaly Averyanov proposed a specific interpretation in the preprint Thirty Names per Sign: the zodiac-ring labels behave like two-letter sigla drawn from closed name lists associated with each sign, encoded through the companion workshop-cipher model.
The paper does not identify the external name list.
It reports that several candidate degree-name source classes fail a pre-declared falsification pipeline.
That is the right place to begin.
The sigla hypothesis is interesting not because it “reads” the labels, but because it converts a vague name-list intuition into a capacity claim and a falsifiable search problem.
Quick Read
- The zodiac labels are a distinct textual population from running Voynich prose.
- The surviving zodiac sequence covers ten signs; Aries and Taurus are split across two folios, producing approximately thirty labels per sign when recombined.
- Averyanov’s 2026 preprint models the labels as two-letter sigla from closed per-sign name lists.
- The paper reports near-zero overlap between ring vocabularies, arguing against a simple internally repeated label list.
- It reports that collapsing labels by proposed homophone or morphological classes still does not create substantial cross-ring overlap.
- A capacity argument is used to claim that the labels carry roughly enough information for two plaintext letters rather than full names or numerals.
- Simulation is then used to estimate the type diversity expected from closed lists of roughly twenty to twenty-five distinctive names sampled across about thirty positions.
- The paper reports a within-ring shared-part pattern interpreted as compatible with independently drawn prefix+suffix encoding.
- Seven published source classes are reported as rejected under the project’s pre-registered crib pipeline.
- No external historical name list has yet been identified strongly enough to turn the model into a decipherment.
- The strongest current claim is structural: bounded zodiac labels may encode closed local inventories differently from running prose.
Why the Zodiac Labels Deserve Their Own Mechanism Test
Most Voynich prose is open-ended.
A paragraph can contain many words.
The reader does not know in advance how many conceptual items it names.
The zodiac ring is different.
The page gives us a finite set of positions.
Each figure has one local label.
The ring provides order and approximate count.
This makes the label problem unusually constrained.
Even if we do not know what is being named, we know the labels behave like an inventory.
That is why a name-list hypothesis is testable in principle.
The First Claim: The Rings Do Not Reuse the Same Vocabulary Much
If the labels meant a small set of repeated categories—“first”, “second”, “male”, “female”, “east”, “west”—we might expect substantial vocabulary overlap between zodiac rings.
The 2026 preprint reports very low inter-ring overlap.
Under its analysis, even collapsing forms by the workshop model’s proposed homophone classes or by morphological cores does not produce the overlap expected from one shared small list.
That supports a model in which each sign carries mostly different local names.
It does not prove those names are personal names.
The closed inventories could theoretically contain:
- degree names;
- star names;
- person names;
- technical categories;
- calendar units;
- other sign-specific labels.
What Is a Siglum?
A siglum is a shortened sign or abbreviation standing for something longer.
Historical manuscripts use abbreviated forms constantly.
One or two letters can represent a name, title, place or technical category if the reader already knows the list from which the abbreviation is drawn.
This is crucial.
A two-letter siglum is not meant to be self-sufficient.
It functions inside a closed codebook.
“JS” can identify one person in a classroom register because the class list is known.
Without the list, “JS” does not tell us whether the person is John Smith, Jane Seah or something else.
This is the exact epistemic position of the zodiac sigla hypothesis.
The Capacity Argument
The preprint asks how much plaintext information a short Voynich label can plausibly carry under the companion cipher model.
Its conclusion is that the capacity is closer to a two-letter siglum than to a full historical name.
This is a useful form of negative evidence.
If a label’s visible capacity is too small to encode a ten-letter name under the proposed mechanism, then full-name readings become less plausible within that model.
But the capacity result is conditional.
Change the cipher model and the plaintext capacity can change.
Therefore:
two-letter capacity is evidence inside the workshop-cipher hypothesis, not an independent measurement of plaintext length.
Why Thirty Positions Do Not Require Thirty Distinct Names
A ring can contain thirty positions while the underlying list contains fewer than thirty distinct types.
Names can repeat.
Some figures may share a category.
The preprint uses simulation to compare observed label diversity with candidate closed-list sizes and reports that a distinctive-word profile of roughly twenty to twenty-five types per thirty positions fits its observed diversity better than several alternatives.
This is a model-selection result.
It is not a historical identification of the list.
Aries and Taurus Make the Count Easy to Misread
The surviving zodiac does not place every sign on one folio.
Aries and Taurus are each split across two panels of about fifteen figures.
Recombine the two halves and the sign-level count returns to approximately thirty.
Gemini is an exception in the surviving material, with twenty-nine in one commonly used count.
This is why all “thirty names” statements should specify whether they refer to:
- figures per physical folio;
- labels per recombined zodiac sign;
- distinct label types;
- expected historical degree systems.
These numbers are related.
They are not interchangeable.
The Shared-Part Result
The preprint reports another structural pattern: some label pairs within the same ring share exactly one of two proposed components above chance.
The authors interpret this as compatible with independently chosen prefix and suffix components in a two-part encoding.
This is interesting because a siglum generated from two independent plaintext initials or two table selections can naturally create partial overlap.
But the interpretation again depends on the segmentation and cipher model.
A different generative grammar could also create shared halves.
The strongest claim is the distributional pattern.
The plaintext interpretation remains conditional.
Why Pharmaceutical Labels Matter as a Control
A good zodiac-label theory should not turn every short Voynich label into the same mechanism.
The preprint compares the ring labels with pharmaceutical labels and reports that the same shared-part signature is absent there.
This is valuable because the two populations are both short labels but likely serve different document roles.
If one mechanism distinguishes them correctly, it gains explanatory power.
The canonical broader owner remains Labels Versus Running Text.
The Crib Pipeline Is More Important Than Any One Rejected List
The paper’s most transferable contribution may be methodological.
A candidate historical name list should not be accepted because three labels happen to resemble three names.
Instead, the candidate list should pass a declared pipeline.
- Does its size fit the ring capacity?
- Does its order align?
- Does the profile of repeated and unique names match?
- Does the proposed encoding predict the observed partial overlaps?
- Are there anachronisms?
- Does the candidate outperform matched alternatives?
The project reports seven published source classes rejected under such controls.
This is useful even if the sigla hypothesis ultimately fails.
It turns the zodiac-label search from resemblance hunting into falsification.
Why a Null Crib Search Is Good Science
The public data-and-code page reports that key-recovery and candidate-list searches have returned honest nulls in important places.
This matters because Voynich research has a long history of finding exactly the name a solver hoped to find.
A pipeline that can say “no” is more valuable than one that always produces a match.
A real crib method must be capable of rejecting the manuscript we want it to confirm.
What Would Turn the Sigla Hypothesis Into a Major Breakthrough?
The missing piece is external.
A period-appropriate, order-aligned list whose entries can be mapped to the surviving ring positions under frozen rules.
The strongest candidate would need to satisfy several demands simultaneously.
- historical availability before or during the manuscript’s production horizon;
- correct number of items per sign;
- correct order;
- correct repeated-name profile;
- correct sex or attribute patterns if those are claimed;
- correct two-letter sigla mapping;
- predictive success on held-out signs.
Without that external anchor, “two-letter sigla” remains a structural mechanism hypothesis.
Why This Does Not Decode the Nymphs
The human figures around the zodiac have tempted researchers to assign identities for generations.
A siglum hypothesis does not entitle us to say which historical person, degree name, star or spirit any one figure represents.
Until the external list is identified, each label remains structurally addressable but semantically unresolved.
This is still progress.
It shrinks the problem from “what could this label mean?” to “what kind of closed list could generate this ring under these constraints?”
The Month Names Do Not Supply the Missing List
The later Romance month names identify the zodiac/month relation for later readers.
They do not identify the original ring labels.
The month annotations are a later layer.
They can confirm Pisces–March, Aries–April and so on without translating the thirty original labels around each sign.
The Biggest Risk: Conditional Evidence Becoming Circular Evidence
The sigla paper depends partly on the workshop-cipher model.
The workshop-cipher model is then strengthened when it explains the zodiac labels.
This can be legitimate if independent observations connect the two.
It becomes circular if every label feature is interpreted through the model and then returned as proof of the same model.
A strong evaluation should therefore separate:
- observations defined without the cipher;
- predictions made by the cipher before inspecting the target;
- post-hoc explanations added after seeing the result.
What the Zodiac Sigla Problem Does Not Prove
- It does not prove the labels are two plaintext letters.
- It does not prove they are personal names.
- It does not prove a particular degree-name tradition.
- It does not identify the historical list.
- It does not recover the cipher key.
- It does not prove the workshop-cipher model.
- It does not turn later month names into original glosses.
- It does not constitute an accepted decipherment.
What It Can Give Us
- A bounded label population ideal for falsification.
- A capacity-based way to distinguish full-name from siglum hypotheses.
- A closed-list model whose expected diversity can be simulated.
- A direct comparison with other label populations.
- A reusable crib pipeline capable of rejecting attractive historical lists.
- A clearer definition of what an external zodiac anchor would have to look like.
Reader Checklist
- What exactly is being counted—figures, labels or distinct types?
- Are split signs recombined correctly?
- How are labels segmented?
- Does the siglum capacity depend on an assumed cipher?
- Does the candidate list fit all signs, not one?
- Is order preserved?
- Are repeated names predicted?
- Are anachronistic candidates rejected?
- Was the crib pipeline frozen before testing the candidate?
- Does the candidate succeed on held-out signs?
- Is a structural fit being described as a reading?
Research Foundations
- Vitaly Averyanov — Thirty Names per Sign (2026 preprint).
- Voynich Studies — Publications and preprint status.
- Voynich Studies — Reproducibility package and crib pipeline.
- The Zodiac Pages.
- Labels Versus Running Text.
The Final Idea
The zodiac labels may be the place where Voynich gives us the most useful kind of constraint:
a finite number of slots, a finite number of labels and a visible order.
That still does not give us the names.
But it tells us what a real name theory must survive.
The sigla hypothesis matters most if it can keep rejecting attractive lists until one historical list finally survives for reasons stronger than resemblance.