Voynich Rare Glyphs: Weirdoes, Ligatures and One-Off Signs

A strange manuscript should not surprise us by containing strange characters.

And yet the rare characters are where Voynich research becomes most vulnerable to surprise.

Most of the manuscript is written with a relatively stable family of recurring signs.

Then a page produces something odd.

A familiar sign with an extra plume.

A compound never seen before.

A form that looks like a table.

An inverted shape.

A character that appears once and then vanishes from history.

Early Voynich researchers needed a way to record this long tail.

Jim Reeds compiled lists of unusual signs not captured adequately by the older standard alphabets.

Jacques Guy referred to many of these as weirdoes while expanding the Frogguy transliteration system.

The nickname is informal.

The research problem is serious.

Every rare sign forces a decision.

New letter?

Rare ligature?

Compound made from known parts?

Personal scribal variant?

Correction?

Accident?

Damage?

Later mark?

The wrong choice can distort an entire theory.

Rare glyphs are dangerous because each carries very little statistical evidence while offering almost unlimited interpretive freedom.

That makes them the exact opposite of the q-series.

q gives us thousands of opportunities to test a rule.

A one-off glyph may give us one.

The rare form is precious.

It is also almost powerless to validate its own meaning.

Quick Read

One-sentence answer: the Voynich Manuscript contains many rare and unusual written forms beyond its common core alphabet, but low frequency makes their status fundamentally uncertain; modern transliteration systems therefore preserve them as distinct observational forms or ligatures while leaving open whether they are true graphemes, compounds, allographs, scribal errors, corrections, damage or later interventions.

Rare forms were already a recognised problem in early modern Voynich transcription work.
Jim Reeds documented dozens of unusual signs outside older standard inventories.
Jacques Guy expanded Frogguy to represent rare forms and the “weirdoes”.
Frogguy’s analytical philosophy can decompose many rare signs into reusable graphical parts.
Extended EVA was explicitly designed to represent rare individual characters and rare ligatures while keeping Basic EVA compact.
Extended EVA distinguishes unusual single characters, ligatures made only from Basic EVA signs, and ligatures that include rare signs.
v101 is more synthetic and preserves many subtle visible variants as separate codes.
Both extended EVA and v101 grew beyond two hundred codes, showing how quickly the observational long tail expands once fine distinctions are preserved.
STA groups rare and common forms into character families partly because intermediate shapes exist and exact boundaries are not always clean.
STA also supports regularisation levels that deliberately collapse some rare distinctions for analysis.
Preserving a rare sign is not the same as declaring it an independent functional character.
One-off signs should carry low semantic confidence because their distribution cannot be tested robustly.
Rare forms are more useful for testing a character-construction model than for assigning isolated translations.
A strong alphabet should explain many rare forms as predicted variants or compounds while preserving the genuinely irreducible exceptions.

The long tail is therefore not where we should begin translating.

It is where we should test whether our theory of the script can survive its hardest edge cases.

Why Rare Characters Become a Transcription Problem Before a Decipherment Problem

Suppose a common Voynich sign occurs five thousand times.

Researchers can measure:

where it occurs;
which signs precede it;
which follow it;
which word families contain it;
how it changes by Currier regime;
whether scribes form it differently.

Now imagine a sign that appears once.

None of those distributions is available.

The first scientific job is therefore preservation.

Record what is visible.

Record uncertainty.

Record whether the form is connected.

Record possible alternative readings.

Do not destroy the evidence by forcing it immediately into a common character.

This is why modern transliteration formats have dedicated mechanisms for rare codes, ligatures and uncertain readings.

The rare sign deserves representation before interpretation.

Jim Reeds’ “Weirdoes” Exposed the Long Tail

Older Voynich alphabets captured the common script reasonably well.

But careful transcription revealed unusual forms that did not fit neatly into those inventories.

Jim Reeds documented many such forms.

Guy’s Frogguy documentation refers to Reeds’ numbered “weirdoes” and adds graphical primitives so those unusual shapes can be represented analytically.

This was an important methodological advance.

Instead of pretending the common alphabet was complete, the transcription system admitted:

the manuscript contains forms our current categories do not explain cleanly.

That is not disorder.

It is good evidence management.

The “Picnic Table” Is a Good Example of Naming Without Meaning

Some rare Voynich forms acquired memorable nicknames.

One was called the “picnic table”.

Others were described as inverted y, squashed z, tent, Chinese hat or similar visual metaphors in early informal transcription work.

These names are useful for conversation.

They carry no semantic authority.

The “picnic table” does not mean table.

A “tent” does not mean shelter.

The nickname only says:

this odd form needs a human-readable handle.

The same safeguard we use for gallows and benches applies even more strongly here.

Visual metaphor must never become decoded meaning.

Extended EVA Separates Three Kinds of Rare Form

The extended EVA framework provides a useful taxonomy.

Rare material can include:

Unusual or rare single characters.
Ligatures made only from Basic EVA characters.
Ligatures that include rare single characters.

This is valuable because not every uncommon visible form expands the alphabet in the same way.

A rare ligature may be built entirely from familiar graphemes.

Its rarity belongs to the combination, not necessarily to the alphabet.

A genuinely unusual simple sign may require a new grapheme candidate.

A ligature involving that rare sign raises both questions at once.

This taxonomy keeps the rare tail from becoming one undifferentiated bucket.

v101 Preserves More Visible Variation

Glen Claston’s v101 alphabet took a comparatively synthetic and fine-grained approach.

Stroke combinations that looked like single signs were often given single codes.

Variants collapsed by earlier alphabets were sometimes distinguished.

This creates a much richer observed inventory.

The benefit is palaeographic fidelity.

The cost is analytical uncertainty.

Some distinctions may be meaningful.

Others may be allographic noise.

A large code inventory therefore says:

we preserved more visible distinctions.

It does not automatically say:

the medieval writing system contained this many independent letters.

Those are different claims.

STA Turns the Long Tail Into Character Families

The modern STA framework takes another useful step.

With extended EVA and v101 each containing more than two hundred codes, simply combining every code into one flat list becomes unwieldy.

STA therefore organises characters into families of similar shape.

This reflects an important palaeographic reality.

Rare signs often do not form clean isolated categories.

Intermediate forms exist.

One sign shades into another.

One scribe adds a flourish.

Another shortens it.

Character families therefore preserve variation without pretending every member is functionally independent.

This is one of the strongest ways to model the rare tail responsibly.

Regularisation Is Not Destruction if the Original Reading Is Preserved

For some analyses, preserving every tiny rare distinction is counterproductive.

A statistical model may become dominated by singleton categories.

STA therefore supports regularisation levels that collapse selected distinctions.

This is scientifically useful if one rule is obeyed:

the high-resolution original classification must remain recoverable.

Then researchers can ask:

Does the result survive when rare variants are collapsed?
Does it depend on treating a one-off form as a separate character?
Which level produces the most robust generalisation?

Regularisation becomes a sensitivity test.

The danger comes only when the lower-resolution version replaces the evidence permanently.

A One-Off Sign Cannot Tell You Its Own Meaning

Suppose a rare glyph occurs beside one plant.

A solver declares:

this glyph means “root”.

How can the claim be tested?

The glyph never appears again.

There is no second context.

No alternative plant.

No syntax.

No distribution.

The picture created the translation, and the same picture is now being used as confirmation.

This is almost pure interpretive freedom.

Rare signs should therefore be among the last characters assigned lexical values, not the first.

Their best role is validation.

Develop the system elsewhere.

Then see whether the rare sign receives a predictable interpretation.

Rare Does Not Mean Important

Human attention overweights novelty.

A common q appears thousands of times and becomes visually boring.

A bizarre one-off mark feels like a secret key.

Statistically, the opposite is often true.

The common sign gives us abundant evidence about system rules.

The one-off sign gives us almost none.

Rare can mean:

special;
accidental;
damaged;
local;
decorative;
obsolete;
scribal.

Frequency alone cannot choose.

The rarity should make us more cautious, not more excited.

Rare Can Still Mean Structurally Important

Caution does not mean irrelevance.

A low-frequency sign can be extremely important if its occurrence is constrained independently.

For example, a rare glyph might:

appear only at paragraph starts;
occur only in one proposed hand;
appear only inside one compound frame;
replace one common glyph under a stable rule;
mark one special document role.

Now rarity has structure.

The sign may be a specialised operator, allograph or grammatical marker.

The key difference is that its importance comes from a repeatable relationship, not from visual weirdness.

Structure rescues rarity from anecdote.

Rare Glyph or Allograph?

One of the first alternatives should always be handwriting variation.

A rare form may be a common character written unusually.

This can happen because of:

scribe;
speed;
pen angle;
line position;
available space;
correction;
fatigue.

Hand classification becomes crucial.

If an odd form appears repeatedly within one proposed hand and substitutes where another hand uses a common glyph, allography strengthens.

If the same unusual form appears across hands in one special context, functional character status becomes more plausible.

Rare-glyph research and the scribe problem therefore belong together.

Rare Glyph or Ligature?

A rare visible form may be built entirely from familiar components.

Extended EVA explicitly preserves such cases as ligatures.

This distinction matters because a rare ligature need not expand the functional alphabet.

It may simply be an uncommon way of joining common graphemes.

The strongest evidence comes from stroke continuity and component recurrence.

If the rare form can be decomposed into elements that independently retain their normal distributional effects, compositionality strengthens.

If the whole form behaves atomically and the parts do not recur meaningfully, a new grapheme becomes more plausible.

Again, the alphabet should be earned by behaviour.

Rare Glyph or Error?

Scribes make mistakes.

A pen slips.

A stroke is doubled.

A correction is written over an earlier form.

A partially erased sign becomes ambiguous.

This means one-off forms can arise without any expansion of the intended script.

Error should not become a convenient wastebasket, however.

An error model should be supported by evidence such as:

overwriting;
incomplete strokes;
correction marks;
nearby intended common forms;
lack of recurrence in equivalent contexts.

“Probably a mistake” is a hypothesis.

The image should support it.

Rare Glyph or Damage?

Parchment changes.

Ink fades.

Stains cross writing.

Creases interrupt strokes.

A damaged common character can become an apparently unique glyph.

This is why rare forms should remain linked to high-resolution manuscript images.

A text file strips away material context.

One high-ASCII code can look perfectly intentional in a corpus even if the original parchment shows abrasion through the middle of an ordinary sign.

Rare characters are where return-to-image discipline matters most.

Rare Glyph or Later Hand?

The manuscript contains later writing layers.

Folio numbers.

Quire numbers.

Month names.

Marginalia.

Ownership marks.

An unusual mark should therefore be assigned to the principal script only after writing stratigraphy is considered.

The three character-like forms in the margin of f1r, for example, have been removed from newer reference transliteration counts because they are treated as later material rather than principal Voynich text.

This is a crucial principle.

Not every character-shaped mark inside the codex belongs to the Voynich alphabet.

Material layer comes before alphabet membership.

Rare Signs Can Distort Entropy

An alphabet with many singleton characters has a different entropy profile from one that collapses them into common families.

This is one reason rare-glyph treatment must be explicit.

If every flourish becomes its own code, alphabet size rises and unigram uncertainty increases.

If rare forms are regularised into common signs, the alphabet shrinks and frequencies concentrate.

Neither is automatically more truthful.

The correct question is whether the distinctions carry reproducible information.

A robust entropy conclusion should survive reasonable rare-character regularisation.

If it disappears only when one-off codes are removed, confidence should fall.

Rare Signs Can Distort Word Counts and Word Families

A single rare glyph inside a common token creates a new word type under exact-string counting.

If the sign is only an allograph, vocabulary diversity has been artificially inflated.

If it is genuinely meaningful, normalising it erases a real distinction.

This is one reason different Voynich transliterations produce materially different counts of unique word types.

Fine-grained alphabets distinguish more token forms.

Coarser alphabets merge them.

Rare-glyph policy therefore changes the apparent size of the vocabulary.

Vocabulary statistics are never independent of palaeographic normalisation.

Rare Signs Are Excellent Tests of a Compositional Alphabet

Suppose the common Voynich script is built from reusable primitives.

Then many rare forms may be unusual combinations of familiar parts.

A good compositional model should predict them.

For example:

common frame + rare payload.

common gallows + unusual flourish.

known minim sequence + rare terminal.

If a model can reconstruct the rare tail from a small component inventory, it gains explanatory compression.

If every rare sign remains an arbitrary exception, the component theory is weaker.

Rare forms therefore test whether visible composition is genuinely productive.

Rare Signs Are Poor Places to Discover a Language

A language identification should be built from repeated structure.

Frequent morphology.

Recurring syntax.

Stable sound or grapheme mappings.

Rare signs should then fall into place.

The reverse workflow is dangerous.

Find one weird sign.

Find one external alphabet with a vaguely similar mark.

Assign the external sound.

Use that sound to identify the language.

This is almost pure resemblance chaining.

The rare sign lacks enough internal evidence to carry that burden.

Common structure should identify the system.

Rare signs should test it.

Rare Signs Are Poor Places to Discover a Cipher Key

The same applies to cryptanalysis.

A one-off cipher symbol could represent:

a rare plaintext letter;
a null;
a nomenclator code;
a punctuation-like marker;
a compound;
a copying error.

Without repetition, frequency analysis provides almost nothing.

A cipher theory should therefore recover the common system first.

Then rare signs can be tested as special codes under the established mechanism.

A rare nomenclator-like symbol is plausible only after a broader cipher explains why most of the manuscript behaves as it does.

One-off complexity should not rescue a failing key.

Rare Signs Can Reveal Production History

Even if a rare glyph never becomes semantically deciphered, it can still be historically useful.

Suppose one unusual form appears only on a set of physical bifolia.

That may support a shared scribe, production batch or exemplar.

Suppose it appears only in later marginal writing.

It becomes a stratigraphic clue.

Suppose the same rare compound appears in two distant sections.

That can connect production practices across page families.

Rare forms therefore have a second life outside decipherment.

They can act as palaeographic fingerprints.

This is often a safer and more productive use of them than assigning meanings.

The Best Rare-Glyph Theory Is Conservative and Reversible

Rare-character handling should preserve two abilities at once.

Distinguish unusual ink when high resolution matters.
Collapse distinctions when testing broader regularities.

This is why reversible mapping between detailed and regularised alphabets is so valuable.

A researcher can ask:

Does my result require every rare distinction?

Does it survive when likely allographs are merged?

Which rare forms genuinely resist decomposition?

The right rare-glyph model should be able to zoom in and out without losing provenance.

That is a stronger scientific design than choosing one alphabet resolution forever.

What Survives the Rare-Glyph Work

The Voynich Manuscript contains a long tail of rare and unusual written forms beyond its common core.
Early researchers explicitly recognised that standard alphabets did not capture all of them.
Jim Reeds catalogued unusual signs, and Guy’s Frogguy system added machinery for representing them analytically.
Extended EVA preserves rare single signs and rare ligatures.
v101 preserves many fine-grained compound and variant distinctions synthetically.
STA groups related rare forms into character families and allows controlled regularisation.
Rare-form preservation is observational, not proof of independent grapheme status.
One-off signs have very low statistical power for semantic assignment.
Hand, position, ligature, error, damage and later writing are serious alternative explanations for rare forms.
Rare glyphs are powerful stress tests for alphabet and production models.
High-resolution image linkage is essential because material damage can masquerade as new characters.

What Does Not Survive as Established Knowledge

Every rare Voynich form is a distinct letter.
Every rare form is merely an allograph.
Every “weirdo” has semantic importance.
A one-off glyph can securely identify a language.
A one-off glyph can securely identify a cipher value.
A visual lookalike in another script proves direct derivation.
All rare signs belong to the principal Voynich hand layer.
Regularising rare forms is always safe.
Preserving every rare distinction means every distinction is functional.

The rare ink survives.

The isolated meaning claims do not.

A Better Rare-Glyph Analysis

Preserve the high-resolution visible form.
Record uncertain readings explicitly.
Separate rare single signs from rare ligatures.
Compare the form with common graphical components.
Check material damage and overwriting.
Check writing layer and later-hand possibility.
Map proposed hand and physical bifolium.
Use character-family models before declaring new letters.
Test analyses at multiple regularisation levels.
Assign semantics only when a broader system predicts the rare form.

What Would Count as a Real Rare-Glyph Breakthrough?

Imagine a compact character-construction model is derived entirely from common forms.

Without fitting the rare signs individually, it predicts that most “weirdoes” should be legal compounds or positional variants.

High-resolution stroke analysis confirms the constructions.

A small residual set remains irreducible and recurs in coherent specialised contexts.

That would dramatically clarify the alphabet.

Or imagine a decipherment derived from common text predicts a value for a rare sign before its occurrence is inspected.

The value fits the rare context and another independently discovered occurrence.

Then the rare glyph finally becomes positive decipherment evidence.

The order matters:

the system should explain the rarity; the rarity should not be used to invent the system.

Primary School: The Strange Letter

Ask a child to write the letter A twenty times.

One A will probably look odd.

Is it a new letter?

Probably not.

Now add one genuinely invented symbol.

How can we tell the difference?

We need more than appearance.

We need repeated use and context.

That is the rare-glyph problem.

Lower Secondary: Preserve First, Simplify Later

Give students ten variants of one handwritten character.

First record every visible difference.

Then group likely variants.

Finally compare the statistics before and after grouping.

The exercise teaches reversible regularisation.

You can simplify for analysis without pretending the underlying evidence never existed.

Upper Secondary: Why One Example Cannot Validate a Dictionary

Give students one invented symbol in one sentence with a picture.

Ask them to guess its meaning.

They can invent many plausible answers.

Now give twenty occurrences across different contexts.

Most guesses disappear.

The student learns why frequency creates falsifiability.

Rare Voynich signs lack that advantage.

JC and Adult Readers: Long-Tail Model Selection

At a higher level, the rare-glyph problem is a classic long-tail trade-off.

A high-complexity model assigns each rare form its own parameter.

It fits perfectly and generalises poorly.

A low-complexity model merges every rare form into a common class.

It generalises well and may erase genuine distinctions.

The ideal model explains most rare forms compositionally, preserves a small evidence-backed residual inventory, and predicts the classification of new rare cases.

That is explanatory compression rather than overfitting.

A Parent and Teacher Guide

Do not confuse unusual with important.
Record rare evidence before simplifying it.
Check allograph, ligature, error and damage before adding a new letter.
Keep later writing separate from principal text.
Use multiple regularisation levels for statistics.
Do not assign meanings from one occurrence.
Make the common system predict the rare case.

The transferable lesson is simple:

rare evidence deserves careful preservation precisely because it is too weak to interpret casually.

Reader Checklist: Before You Make a Rare Glyph a New Character

How many times does the form occur?
Is the reading certain?
Could damage explain it?
Could overwriting or correction explain it?
Does it belong to the principal writing layer?
Could it be a ligature of common components?
Could it be an allograph associated with one hand?
Does line or paragraph position explain the variant?
Which extended alphabet records it?
Does STA place it in a broader character family?
Does a regularised corpus preserve the main result without it?
Can a broader decipherment predict its function without using the rare occurrence to invent the key?

Frequently Asked Questions

What are Voynich “weirdoes”?

It is an informal historical nickname for unusual Voynich forms catalogued beyond the common character inventory, especially in work associated with Jim Reeds and Jacques Guy.

Are rare glyphs real characters?

Some may be. Others may be ligatures, compounds, allographs, errors, damaged signs or later marks. Low frequency often prevents a decisive classification.

What is the “picnic table”?

It is a visual nickname used in older Voynich discussion for one unusual sign. The nickname has no decoded semantic meaning.

Why does Extended EVA exist?

Basic EVA handles the common script efficiently. Extended EVA adds codes and conventions so rare characters and ligatures can be preserved in complete transliterations.

Why does v101 have so many characters?

Its design is comparatively synthetic and fine-grained, preserving many compound forms and visible variants that other alphabets decompose or merge.

What does STA do with rare glyphs?

It incorporates distinctions from major alphabets, groups related forms into character families and allows controlled regularisation so researchers can test whether results depend on fine rare-character distinctions.

Can a rare glyph identify the Voynich language?

Not securely by itself. A one-off sign provides too little internal evidence. Language identification should come from repeatable mappings across common text and then predict the rare forms.

What is the strongest current conclusion?

The rare signs are genuine observational evidence of a complex handwriting system, but their functional status remains heterogeneous and uncertain. They are best used as stress tests of broader character, scribal and decipherment models.

Research and Further Reading

The Final Idea

The rare glyphs are the manuscript’s temptation.

They look special.

We want them to be keys.

We want the strangest sign to unlock the strangest book.

Science asks for the opposite temperament.

Preserve the odd sign carefully.

Lower confidence because it is rare.

Test allograph, ligature, damage and error.

Build the system from repeated evidence.

Then return to the strange sign.

If the system predicts it, the rare glyph becomes powerful.

If the system needs the rare glyph to rescue itself, the rare glyph becomes a warning.

The hardest character should be where a decipherment proves its discipline, not where it escapes it.

Continue Through the Voynich Research Map

This article is one specialist node in eduKateSG’s larger Voynich research library. Return to the canonical master to see the physical, visual, textual and historical evidence in one continuous argument.

Structure is evidence; structure is not translation. The master keeps resemblance, statistical structure, historical possibility and decipherment claims at separate evidentiary levels.

Next Best Routes

The Alphabet Problem — decide whether the rare tail genuinely expands the inventory or mostly records variants and compounds.
Abbreviation or Alphabet? — test whether uncommon forms belong to a larger scribal compression system rather than a larger phonetic alphabet.
The Bench Characters — compare one-off complexity with a frequent compositional family whose parts can be studied repeatedly.

Deep bridge: What Is Robustness? provides the right test: a useful character model should keep its conclusions when rare forms are regularised, isolated, or reclassified under reasonable alternatives.

Voynich | Everything eduKate Knows and Tested | Rare Glyphs: Weirdoes, Ligatures and the One-Off Signs