Voynich Herbal Pages: Why the Plants Resist Identification

The Voynich Manuscript becomes most dangerous when it looks almost familiar.

A leaf curves the way you remember a leaf curving.

A flower resembles something from a garden.

A root seems unmistakable for three seconds.

Then you look again.

The leaf belongs to the wrong stem.

The flower has the wrong number of parts.

The root behaves like no root you know.

Another researcher identifies the same page as a completely different species.

A third person says the plant is composite.

A fourth says it is copied badly from an older herbal.

A fifth says the strange parts are the entire point.

This is why the Voynich “herbal” pages are much more interesting than a botanical guessing game.

The real problem is not:

Which modern plant does this picture resemble?

It is:

What kind of visual information system produces a plant-like image with these particular choices, on this particular page, beside this particular kind of text?

That question is slower.

It is also much closer to the manuscript.

Quick Read

One-sentence answer: the Voynich herbal pages are a large family of plant-dominated folios whose image-and-text architecture resembles medieval herbal practice in broad ways, but most individual plant identifications, exact functions and textual meanings remain unresolved because resemblance, copying, stylisation, diagrammatic emphasis and modern taxonomy can all produce misleading certainty.

“Herbal” is a useful modern descriptive label for plant-dominated pages; it is not a decoded medieval section title.
Many folios present one large plant-like figure with substantial unidentified text arranged around or beside it.
The drawings contain recognisable botanical components—roots, stems, leaves, flowers or flower-like structures—but many combinations resist secure modern species identification.
A medieval plant image did not have to function like a modern field-guide photograph.
Images could be copied from exemplars, simplified, exaggerated, recombined, transformed through generations of copying or designed to emphasise useful features rather than natural appearance.
A modern species match should use multiple independent botanical features, not one memorable resemblance.
A mismatch cannot automatically be rescued by declaring the image “composite”; a composite hypothesis needs rules and predictions of its own.
Roots are especially vulnerable to pareidolia because irregular branching forms can resemble animals, tools, letters and other familiar objects.
The Carrara Herbal, British Library Egerton MS 2020, proves that highly resolved medicinal plant illustration existed in a sophisticated northern Italian manuscript environment around 1400. It does not prove Voynich provenance.
Masson 116 and Sloane MS 4016 are valuable because Sloane 4016 is catalogued as a copy of Masson 116, giving us a known case of visual transmission and transformation.
Egerton MS 747 shows that plant knowledge could coexist in one codex with antidotaries, doses, substitutions, weights, measures, synonyms and a lunar-calendar fragment.
Currier A and Currier B both occur within broadly plant-dominated material, showing that one visual family does not correspond automatically to one textual regime.
Detached plant components in later “pharmaceutical” pages may connect to whole-plant pages, but internal resemblance does not by itself prove identical plant identity.
The strongest herbal research separates morphology, page role, textual behaviour, physical construction and historical comparison before attempting semantic naming.

The plants are not failing because we have not stared at them hard enough.

They are telling us that identification is a multi-layer problem.

First, What Do We Mean by “Herbal”?

A herbal is, broadly, a work concerned with plants and their names, properties, uses or related knowledge.

But the category stretches across centuries and manuscript traditions.

Some herbals organise substances alphabetically.

Some derive from classical or Arabic authorities.

Some emphasise medicinal use.

Some become picture books with short captions.

Some contain long prose.

Some are copied from older exemplars whose images have already travelled through several generations.

So when scholars call part of the Voynich Manuscript “herbal”, the safest meaning is visual and comparative:

these pages prominently depict plant-like forms in a page architecture that invites comparison with herbal traditions.

That sentence does not yet tell us whether the text gives names, remedies, properties, agricultural instructions, mnemonic information or something else.

The section label locates the reader.

It should not silently perform the decipherment.

The Page Architecture Is Already Information

Many Voynich herbal folios share a striking broad architecture.

One large plant dominates the visual field.

Text occupies the available space around it.

Lines may wrap beside leaves or stems.

The plant and the writing are not laid out as two completely separate pages accidentally bound together.

They share one information surface.

This tells us something before translation.

The image mattered enough to control page space.

The text mattered enough to be fitted around the image.

Whatever the intended function, the user was probably expected to encounter image and writing together.

That does not prove that the text describes the plant directly.

But it makes a relationship between page image and page text a serious hypothesis rather than an arbitrary association.

A Plant Has Parts; an Identification Needs Several of Them

Imagine trying to identify a real plant from one feature.

“The leaf is heart-shaped.”

Many plants have heart-shaped leaves.

“The flower is yellow.”

Thousands of flowers are yellow.

“The root is bulbous.”

Again, not enough.

Botanical identification gains strength when several diagnostic features converge:

leaf arrangement;
leaf margin;
venation;
stem structure;
flower symmetry;
number and arrangement of floral parts;
fruit or seed structure;
root or underground organ;
growth habit;
geographic and ecological plausibility.

The Voynich drawings complicate nearly every one of these.

Some details are stylised.

Some are unclear.

Scale is unstable.

Colour may not be naturalistic.

Roots can be spectacularly strange.

Therefore a convincing identification should survive more than one component.

One matching leaf is a clue.
A coherent plant morphology is evidence.

Modern Taxonomy Is Not the Only Historical Ontology

Modern readers naturally want a Linnaean answer.

Genus.

Species.

One picture, one taxon.

The fifteenth-century maker did not live inside modern botanical taxonomy.

Plants could be organised through names inherited from authorities, medicinal substances, local synonyms, usable parts, perceived qualities, preparations or practical associations.

A medieval “plant entry” might therefore correspond imperfectly to what a modern botanist calls a species.

A name might refer to a drug substance.

One substance could have several regional names.

Several plants might be treated as substitutes.

A copied tradition might perpetuate an old image even after direct botanical observation weakened.

This does not make identification impossible.

It means the target may not be “modern species portrait” in the first place.

A Medieval Plant Image Could Be Copied From a Picture, Not a Plant

This changes everything.

Suppose Artist A observes a living plant.

Artist B copies Artist A’s picture fifty years later.

Artist C copies B.

Artist D copies C but has less skill.

The final image is not a direct observation of a plant.

It is an observation of an observation of an observation.

Features can drift.

Curves simplify.

Leaf counts change.

Roots become decorative.

Colour is inherited by convention rather than observation.

This is why manuscript transmission matters as much as botany.

The thing being identified may be a place in an image tradition rather than a direct portrait of a specimen.

Masson 116 and Sloane 4016 Show What Real Visual Transmission Can Look Like

This is why MS Masson 116 and British Library Sloane MS 4016 are such valuable controls.

Biblissima describes Masson 116 as a Tractatus de herbis manuscript whose illustrations include plants and animals.

The British Library describes Sloane MS 4016 as an Italian herbal from the North Italian group and as a copy of Masson 116, dating it around 1440.

That relationship gives us something Voynich comparison usually lacks.

A known lineage.

We can examine how copied plant images retain identity while changing appearance.

We can observe what features survive.

What simplifies.

What is exaggerated.

What the copyist seems to misunderstand.

This creates a much higher standard for proposed Voynich source relationships.

If somebody claims a Voynich plant was copied from another herbal tradition, the strongest case should involve a family of systematic correspondences, not one leaf whose silhouette feels familiar.

Copying Can Preserve Identity While Destroying Naturalism

This is an important distinction.

An image can become botanically worse and historically more recognisable.

If a community knows the convention, a simplified image can still retrieve the intended name.

Think of a subway symbol.

It does not resemble a physical train station realistically.

It works because the user knows the code.

A historical plant-image tradition can also acquire conventional features.

This suggests a useful question for Voynich:

Are some strange features failures of observation, or are they stable features of a visual code whose users already knew what mattered?

We cannot answer that globally yet.

But it is a better question than assuming every distortion is incompetence.

The Carrara Herbal Proves High-Resolution Plant Illustration Was Possible

British Library Egerton MS 2020, the Carrara Herbal, is particularly useful because it sits close to the Voynich material window and comes from a highly sophisticated manuscript environment.

The British Library describes it as an Italian translation of a work associated with Serapion the Younger, decorated for the Carraresi family of Padua and containing numerous coloured plant miniatures.

Its plants can be astonishingly specific.

The catalogue identifies examples such as citron, violet, melon, cucumber, lettuce, peach and other plants.

Why does this matter?

Because it weakens one lazy explanation:

Voynich plants look strange because medieval people simply could not draw plants accurately.

Some medieval artists clearly could.

Therefore Voynich stylisation may reflect many possible causes:

different artistic skill;
different exemplars;
different purpose;
copying degradation;
diagrammatic emphasis;
intentional transformation;
or a combination.

The Carrara Herbal does not tell us which cause applies.

It tells us that inability is not the only historically available explanation.

And Again: Carrara Is a Control, Not a Birth Certificate

The danger of a good comparison is that it becomes too good.

The Carrara Herbal has the right broad period.

It has medicine.

It has plants.

It has Padua.

It has sophisticated visual organisation.

It is therefore tempting to make the final leap:

Voynich came from the same environment.

That is still not established.

The comparison demonstrates capability and historical plausibility.

It does not create a custody chain.

This distinction is important enough to repeat because once provenance contaminates visual interpretation, every later resemblance starts “confirming” the assumed origin.

Good comparison expands the search space intelligently.

It does not close it dishonestly.

Egerton MS 747 Shows Why “Herbal” Can Be Part of a Larger Working Book

British Library Egerton MS 747 is one of the most useful reminders that medieval knowledge did not live in our modern school-subject boxes.

Its contents include a Tractatus de herbis, additional plant images, a lunar-calendar forecasting fragment, the Antidotarium Nicolai, material on doses, ingredient substitutions, weights and measures, synonyms for plants and medical ingredients, and other additions.

That is not one modern discipline.

It is a practical knowledge environment.

This matters for Voynich because the manuscript also contains broad visual families that modern cataloguers describe separately.

Plants.

Zodiac imagery.

Human figures.

Containers.

Starred entries.

Egerton 747 shows that some combinations of herbal, pharmaceutical and calendrical knowledge are historically ordinary enough to take seriously.

But coexistence is not a decoded pipeline.

A known codex can contain related modules without every module being one sequential procedure.

The Herbal Pages Contain Both Currier A and Currier B

This is one of the most important things the text tells us about the images.

Plant-dominated pages do not all belong to one textual regime.

Currier A and Currier B both occur within the broad herbal visual family.

This immediately defeats the simplest equation:

same kind of picture = same kind of text.

Something else changes.

Perhaps topic at a finer level.

Perhaps source exemplar.

Perhaps writing convention.

Perhaps cipher state.

Perhaps production phase.

Perhaps several of these.

The image family is real.

The textual diversity inside it is real too.

A strong herbal theory has to explain both.

The Bifolium Makes the Herbal Problem Even More Interesting

Currier observed that herbal bifolia tended to preserve A or B consistently across both halves of the physical sheet.

This is a remarkable connection between text and manufacture.

The broad visual topic remains plant-like.

The textual regime changes between physical production units.

That suggests the hidden variable may live partly in workflow.

Perhaps one batch of source material was copied onto one set of sheets.

Perhaps different writers or writing states occupied different bifolia.

Perhaps the intended herbal taxonomy itself was organised in physical batches.

None is proven.

But the result teaches us that plant identification alone cannot explain the herbal section.

The book-making process is part of the problem.

The Root Is Not Just the Bottom of the Plant

Voynich roots often carry disproportionate visual drama.

They can be huge.

Forked.

Bulbous.

Geometric.

Animal-like.

Tool-like.

Sometimes almost architectural.

This has generated countless symbolic identifications.

The root may indeed carry more information than naturalistic botany requires.

But several explanations remain available:

the medicinally important underground part is being emphasised;
the image has drifted through copying;
the root contains a mnemonic convention;
the root encodes a category;
the artist has stylised freely;
the apparent resemblance is pareidolia.

A symbolic-root theory becomes much stronger if the same form recurs in the same role across several independent plants.

Without recurrence, one strange root can tell almost any story.

Roots Are a Pareidolia Machine

Irregular branching geometry is perfect material for human pattern-recognition.

One person sees a dragon.

Another sees scissors.

Another sees a zodiac animal.

Another sees a letter.

The shape has not changed.

The observer has.

This is why a visual resemblance needs a denominator.

How many alternative objects could the same root resemble?

How often do similarly complex roots accidentally resemble familiar things?

Did the researcher decide what to look for before seeing the root, or after?

A striking match with a vast search space can be less impressive than it feels.

Leaves Are Safer Than Roots—but Not Safe

Leaves offer botanical structure that can be compared more systematically.

Margin.

Shape.

Attachment.

Arrangement.

Venation.

These can narrow a candidate set.

But a copied leaf can be simplified.

A lobed leaf can become serrated.

A compound leaf can be drawn as several independent leaves.

Venation can be decorative rather than diagnostic.

So leaf morphology is valuable only when the rest of the plant does not strongly contradict it.

A good identification is a coalition of features.

Not a dictatorship of the most familiar leaf.

Flowers Can Be Diagnostic—and Stylisation Can Destroy That Advantage

In modern botany, flowers can provide extremely useful taxonomic characters.

Number of petals.

Symmetry.

Arrangement.

Inflorescence.

But a medieval illustrator may not preserve exactly those features.

A flower may be drawn as a generic star.

Several blossoms may be compressed.

Colour may be conventional.

A copyist may reproduce a stylised floral motif rather than a botanical organ.

The identification problem therefore requires historical image literacy in addition to plant science.

Colour Is a Weak Species Identifier Unless the Manuscript Proves It Is Stable

A blue flower invites a blue-flowered species.

That feels obvious.

Historical pigments make it less obvious.

Colour can be limited by available materials.

It can be added in a later production stage.

It can be decorative.

It can classify rather than imitate.

It can age.

Therefore colour becomes strong taxonomic evidence only if its naturalistic reliability can be demonstrated across known cases.

Until then, treat pigment as another variable:

interesting when correlated, dangerous when isolated.

Scale Can Turn Useful Parts Into Monsters

A root can occupy half a page.

A flower can shrink.

A stem can bend unnaturally to avoid text.

Modern viewers may interpret this as botanical absurdity.

But diagrams often change scale to preserve information.

A subway map expands central stations.

An anatomy diagram enlarges a tiny structure.

A botanical plate may isolate a useful organ.

If the Voynich herbal pages are partly diagrammatic, disproportion might be functional.

That hypothesis becomes stronger if the enlarged components correlate with other variables such as text, repeated motifs or detached fragments elsewhere.

Scale is evidence.

Literal scale is an assumption.

The Composite-Plant Hypothesis Is Useful Only If It Can Lose

“Composite plant” is an attractive explanation because it solves mismatch immediately.

The leaf matches Plant A.

The flower matches Plant B.

The root matches Symbol C.

Done.

Except that a theory able to explain every mismatch cannot be tested.

A serious composite model should specify a compositional grammar.

Which parts are allowed to come from different plants?
Why those parts?
Do the same donor components recur elsewhere?
Does the composite encode medicinal substitution, taxonomy, mnemonic information or another stable job?
Do textual features correlate with the proposed components?
Can the model predict a composite on an unseen folio?

If the answer is simply “any strange part may come from anywhere”, the hypothesis has become permission rather than explanation.

A Badly Copied Real Plant and a Deliberately Composite Plant Can Look Similar

This is one of the hardest visual-identifiability problems.

Suppose the final image combines incompatible-looking features.

There are at least two broad stories.

Story 1: the image began as one real plant and became distorted through copying.

Story 2: the image intentionally combines features from multiple sources.

Both can produce the same final mismatch.

How do we distinguish them?

We need an independent discriminator.

A transmission series showing how one source image degrades.
Repeated modular components reused across different Voynich plants.
Textual labels that track components rather than whole plants.
A known contemporary visual convention for constructing composites.

Without that extra evidence, both mechanisms remain live.

This is why visual strangeness does not name its own cause.

The “One Plant, One Word” Dream Is Too Easy

An obvious decoding fantasy goes like this.

Identify the picture.

Find the nearby label.

Declare the label to be the plant’s name.

Repeat.

Soon a dictionary appears.

The problem is that every arrow contains an assumption.

The picture has been correctly identified.
The nearby token is actually a label for the whole plant.
The token encodes a name rather than a property, instruction or index.
The visible space boundaries define the correct textual unit.
The same plant uses the same name everywhere.
The writing system maps to language in the expected way.

If the first plant identification is wrong, the proposed dictionary poisons every later reading.

This is why image-based decipherment needs held-out prediction.

Do not let the picture write the translation and then use the picture to confirm it.

Labels and Paragraphs May Be Doing Different Jobs

Some Voynich pages contain isolated short textual units, while many herbal pages contain longer running text.

A short label beside an image component may have a different function from a token buried inside prose-like text.

That sounds obvious.

It has major analytical consequences.

A label could be:

a name;
a code;
a category;
a quantity;
a cross-reference;
a short instruction.

A paragraph can carry relationships that labels do not.

If the same word-like form appears in both environments, its role may still differ.

Therefore herbal research should preserve text type as a variable rather than merging every token into one giant vocabulary bucket.

Whole Plants and Detached Plant Parts May Form an Internal Crosswalk

Later pages conventionally called pharmaceutical contain smaller plant fragments alongside repeated vessel-like forms.

This creates one of the most interesting internal comparison opportunities in the manuscript.

Do some detached roots, leaves or stems resemble components from earlier whole-plant pages?

If yes, that could support several relationships.

The same plant appears twice.
The same medicinal part appears separately.
The same visual template is reused for a class.
The same artist uses habitual forms.
Different plants share a generic component convention.

The internal nature of the comparison is useful because we do not need to prove an external transmission route.

But identity is still not automatic.

A strong internal match should combine several independent features:

geometry + attachment + colour + textual context + page role.

The more dimensions align, the more interesting the proposed crosswalk becomes.

The Same Shape Can Mean Same Object—or Same Class

This ambiguity sits at the heart of internal image matching.

Imagine several modern maps use the same blue wave symbol.

It does not mean the same river appears on every map.

It may simply mean “water”.

Likewise, a repeated Voynich root form could mean:

same plant;
same type of root;
same medicinal category;
same artist template;
same process state;
or nothing semantic at all.

Recurrence proves reusability.

It does not by itself tell us what level is being reused.

New World Plant Claims Need an Exceptionally High Bar

Some Voynich plant identifications have been used to argue for contact with plants associated with the Americas.

Such a claim is not merely botanical.

It has chronological and historical consequences.

The parchment samples fall in the early fifteenth century.

Therefore a confident identification of a plant whose European knowledge depends on later transatlantic exchange would require us to revisit either the plant identification, the production chronology, the transmission story or several assumptions at once.

That does not make a surprising identification impossible by definition.

It makes the evidentiary burden much higher.

A strong case would need multiple diagnostic botanical features and an historically plausible route.

One vaguely familiar flower cannot carry a rewriting of global chronology.

Geography Is a Constraint After Identification, Not a Substitute for It

A candidate plant can be botanically plausible and geographically implausible.

Or geographically plausible and botanically weak.

Both dimensions matter.

But geography should not be used backwards.

If a researcher already believes the manuscript came from Padua, they may preferentially search for Paduan or Mediterranean plants.

Then those plant identifications appear to confirm Padua.

The loop is circular.

A stronger sequence is:

describe morphology without provenance assumptions;
rank botanical candidates;
then examine geographic and historical plausibility;
finally ask whether the result contributes independent provenance evidence.

The order protects the conclusion from the premise.

The Plants May Encode Function Better Than Species

Here is a possibility worth keeping alive without turning it into fact.

What if the strange drawings are optimised for use rather than taxonomic portraiture?

A medicinally important root might be enlarged.

A diagnostic leaf may be simplified.

A flower may merely identify a broad class.

Several stages of growth may be compressed onto one page.

The image may be a retrieval interface for someone who already knows the plant.

This would explain why modern species identification is difficult while internal users could still find the image useful.

But a function-first model needs predictions.

Do enlarged roots correlate with later detached roots?

Do text patterns differ by emphasised organ?

Do known practical herbals use comparable emphasis?

Without those correspondences, “functional diagram” remains a possibility rather than an explanation.

The Page Could Be an Entry, Not a Portrait

This conceptual change is useful.

Instead of saying:

This page is a picture of Plant X.

say:

This page appears to be an entry organised around a plant-like visual object.

An entry can contain more than identity.

Name.

Synonym.

Property.

Preparation.

Use.

Substitution.

Warning.

Source authority.

Cross-reference.

This wider information architecture is much closer to what known historical herbals and pharmacological manuscripts can do.

And it explains why translating one token as the plant name would not automatically unlock the surrounding paragraph.

Synonyms Matter More Than Modern Readers Expect

Egerton MS 747 explicitly contains synonym material for plants and medical ingredients.

This reminds us that naming itself was a practical problem.

The same substance can travel through languages and regions under different names.

A practitioner needs to know that Name A, Name B and Name C may refer to the same usable material.

If Voynich contains a comparable concern, repeated or related labels might reflect synonym systems rather than one-to-one naming.

We are not claiming that they do.

The comparison simply expands the plausible job of text around a plant.

A plant page can be about the difficulty of naming as much as about the plant itself.

Substitution Changes What “Same Plant” Could Mean

Egerton 747 also contains a Quid pro quo list—substitutions for use when an ingredient is unavailable.

This is a small historical fact with a large conceptual consequence.

A practical medical system can group different substances because they perform similar jobs.

So visual or textual association does not always imply botanical identity.

Two plant components might be linked because one substitutes for another.

Two names might co-occur because they are alternatives.

A page family might classify remedies by use rather than taxonomy.

This is another reason modern one-species-one-entry assumptions can be too narrow for historical pharmacological material.

Again, possibility is not identification.

But known manuscripts teach us what kinds of alternative organisation existed.

The Clean Plant → Remedy Pipeline Did Not Survive as a Proven Architecture

The most tempting whole-manuscript story begins here.

First identify a plant.

Then detach the useful part.

Then place it in a vessel.

Then apply it to a body.

Then time the treatment astrologically.

Then record the final recipe.

It is elegant.

It is historically conceivable.

And it did not survive strongly enough to be promoted into established architecture.

The current page order is not secure enough to carry the sequence by itself.

The visual section labels are descriptive rather than decoded.

Container types do not provide an established process legend.

Zodiac imagery does not independently prove a master timing layer.

Currier regimes do not align as a simple progression through those stages.

The correct response is not to throw away every local connection.

It is to decompose the grand story into smaller questions that can survive independently.

Smaller Questions Are Stronger

Instead of asking whether the entire manuscript is one herbal-medical machine, ask:

Do detached plant parts systematically match whole-plant components?
Do matched components share related labels?
Does root emphasis correlate with later plant-fragment pages?
Do A and B herbal pages differ visually in measurable ways?
Do certain page-layout families correspond to textual subgroups?
Do container-like forms correlate with particular classes of plant parts?
Does a proposed botanical family predict recurring textual forms?

Each question can fail without destroying the others.

That is what makes them scientifically useful.

A good research programme is often a large beautiful theory broken into pieces small enough to be wrong.

What Survived From the Herbal Work

The plant-dominated pages form a genuine major visual family.
Image and text are deliberately coordinated on the page.
Many plant-like drawings contain systematic component structure rather than arbitrary scribbling.
Some components recur internally and deserve controlled matching.
Roots, leaves, flowers, colour and scale should be analysed separately before being recombined into an identification.
Medieval herbal comparison greatly improves interpretation of possible image functions.
Known copying relationships show that visual identity can survive substantial transformation.
The Carrara Herbal demonstrates that high-resolution naturalistic depiction was possible near the relevant historical period.
Currier A/B differences inside the herbal family prove the broad visual category is not textually uniform.
Physical bifolia add a production dimension that species identification alone cannot explain.
Whole-plant and detached-fragment relationships remain a productive internal research lane when treated probabilistically rather than as decoded identity.

Those are strong results.

None requires us to pretend we know the Latin name of every page.

What Did Not Survive as Established Knowledge

A complete reliable modern species catalogue for the Voynich herbal pages.
A one-to-one plant-name dictionary derived from adjacent tokens.
A universal symbolic-root code.
A rule that every strange plant must be composite.
A rule that every strange plant is merely a badly copied real specimen.
A direct Carrara or Paduan provenance derived from visual similarity.
A proven whole-plant → fragment → vessel process sequence.
A botanical identification strong enough by itself to establish a New World production history.
A single semantic explanation for why both Currier A and B occur in the herbal visual family.

Keeping those failures visible is not pessimism.

It stops the next researcher from rebuilding the same certainty on the same weak foundation.

A Better Plant-Identification Ladder

1. Describe

Record visible morphology without naming the plant.

2. Segment

Separate root, stem, leaf, flower, fruit-like forms, colour and page placement.

3. Search internally

Look for recurring components elsewhere in Voynich before importing external candidates.

4. Generate several botanical candidates

Do not allow the first familiar silhouette to monopolise the search.

5. Count agreements and contradictions

A candidate should explain the whole morphology better than alternatives.

6. Compare manuscript traditions

Ask whether the image could derive from a known visual convention rather than direct observation.

7. Add textual and physical context

What Currier regime, page layout, bifolium and label environment surround the image?

8. Add geography and chronology last

Use them to test the candidate rather than to generate the answer circularly.

This ladder does not guarantee an identification.

It makes false confidence more expensive.

What Would Make a Voynich Plant Identification Strong?

Several independent diagnostic botanical features agree.
Major contradictory features are few and explicitly explained.
The candidate was not selected only because one component looked familiar.
The identification remains plausible under known medieval copying distortions.
Its geographic history is compatible with the manuscript’s material period or has a documented transmission route.
The same plant or component recurs elsewhere in a consistent way.
Associated textual behaviour predicts rather than merely follows the visual identification.
Independent researchers can apply the same criteria and rank the same candidate highly.
The theory distinguishes the candidate from close alternatives.
The identification improves explanation of the manuscript rather than creating new ad hoc exceptions everywhere else.

A strong identification should make the manuscript simpler in a disciplined way.

Not simpler because we stopped counting contradictions.

What Would Count as a Real Herbal Breakthrough?

Imagine a set of twenty plant pages independently identified from morphology.

Now imagine their nearby labels are decoded under one stable textual rule.

The rule produces historically appropriate plant names or synonyms.

Then the same rule is applied to ten unseen plant pages.

It predicts candidates before the images are inspected in detail.

Those predictions match morphology.

Related detached fragments recur later with related labels.

Currier differences are accounted for by a principled rule.

Independent researchers reproduce the result.

That would be a breakthrough.

Notice how different it is from:

This leaf looks like basil and I can make the nearby token sound vaguely like a word for basil.

The first system predicts.

The second recognises after the fact.

Primary School: Describe the Plant Without Naming It

Give a child any strange plant drawing.

Ask them to write only what they can point to.

four large leaves;
two small leaves;
a thick stem;
three flower-like circles;
a forked root;
blue colour at the top.

Then ask:

Now what do you think it might be?

The learner immediately experiences the difference between description and identification.

Then ask for three possible answers instead of one.

That small change teaches intellectual flexibility without sacrificing evidence.

Lower Secondary: Build a Morphology Table Before Searching

For Secondary 1 and Secondary 2, make students record features before they are allowed to search for candidates.

leaf shape;
leaf arrangement;
margin;
stem;
flower;
root;
colour;
scale;
unusual features.

Why before searching?

Because the first search result contaminates what the eye notices next.

If a website says “this is Plant X”, the learner begins seeing the matching features and forgiving the mismatches.

Recording first preserves an independent observation layer.

That is a scientific habit disguised as art history.

Upper Secondary: Compare Identification Mechanisms, Not Just Species

By Secondary 3 and Secondary 4, give students one strange plant and four mechanism hypotheses:

direct naturalistic observation;
badly degraded copy;
deliberate composite;
functional diagram.

Then ask what each mechanism predicts.

A degraded-copy model predicts a recoverable source family.

A composite model predicts modular reuse.

A functional-diagram model predicts correlations between exaggerated parts and textual or later visual use.

The student moves from asking:

What plant is it?

to:

What process made it look like this?

That is a much higher-order question.

JC and Adult Readers: Treat Plant Identity as a Latent Variable

At a higher level, the true plant identity—if a single true modern identity exists—is hidden.

We observe a transformed representation.

The transformation can contain noise from:

copying;
stylisation;
scale;
pigment;
artist skill;
historical taxonomy;
intended function;
damage.

The inference problem is therefore:

Which hidden source object and which transformation process together best explain the image?

This is much closer to inverse modelling than simple visual matching.

It also explains why two researchers can look at the same image and choose different plants while both finding genuine similarities.

The missing variable may be the transformation process itself.

A Parent and Teacher Guide

The herbal pages can teach an unusually transferable reasoning routine.

Observe first. Record what is visible before searching.
Generate alternatives. Never reward the first familiar match simply for arriving first.
Count contradictions. A good theory must explain what does not fit.
Separate object from representation. A strange drawing may be a transformed image of an ordinary thing.
Check the historical world. Ask what manuscript makers of the period could actually do.
Keep comparison bounded. Similarity does not prove provenance.
Ask for prediction. What should we see on another page if this interpretation is right?

This routine works for far more than medieval plants.

It works for graphs.

News photographs.

Scientific diagrams.

Historical claims.

And any situation in which recognition feels faster than evidence.

Reader Checklist: Before You Believe a Voynich Plant Identification

Was the plant described before it was named?
How many independent botanical features match?
Which important features contradict the candidate?
Is the identification driven mainly by one leaf, flower or root resemblance?
Could copying degradation explain the mismatch?
Could deliberate stylisation explain it?
If “composite” is proposed, what rules determine the components?
Does the same component recur elsewhere in Voynich?
Does associated text provide independent support or was the image used to invent the reading?
What Currier regime and physical bifolium contain the page?
Does a known manuscript tradition provide a plausible transmission analogue?
Is a comparator being misused as provenance?
Is the proposed plant historically and geographically plausible for the production window?
Would the theory survive an unseen page?
What observation would make the identification fail?

The last question is the one most likely to protect you from a beautiful mistake.

Frequently Asked Questions

Are the Voynich herbal pages really herbals?

They are reasonably described as herbal-like because large plant images dominate many pages and invite comparison with medieval herbal traditions. The exact medieval category and function remain undeciphered.

Can the plants be identified?

Some individual proposals are plausible, but there is no broadly accepted complete identification set. Many drawings combine familiar and anomalous features.

Are the plants imaginary?

Not established. The images may include real plants represented through stylisation, copying, diagrammatic transformation or possible compositing. “Imaginary” is only one explanation family.

Are they composite plants?

Some may be, but the hypothesis should not be used as a universal rescue for mismatching features. A strong composite model requires repeatable rules and internal predictions.

Why are the roots so strange?

Possible reasons include functional emphasis, copying transformation, stylisation, symbolic or classificatory use and ordinary drawing variation. Their ambiguity also makes them particularly vulnerable to pareidolia.

Does the Carrara Herbal decode Voynich plants?

No. It demonstrates that sophisticated, highly resolved medicinal plant illustration existed in a relevant historical environment. It is a comparator, not a decoding key or provenance certificate.

Why do Masson 116 and Sloane 4016 matter?

Because Sloane 4016 is catalogued as a copy of Masson 116. Together they show what a supported visual transmission relationship looks like and how copied images can change while retaining lineage.

Do all herbal pages use the same Voynich text type?

No. Currier A and Currier B both occur in broadly plant-dominated material. The herbal visual family is not textually uniform.

Do detached plant parts prove recipes?

No. They create a plausible internal relationship with whole-plant pages, but detached components could support identification, classification, storage, ingredient reference, preparation or other functions.

What is the strongest conclusion?

The herbal pages are a deliberately organised plant-centred information system in which image, text and physical production interact. The exact plant identities and intended semantic function remain only partially recoverable.

Research and Further Reading

The Final Idea

The Voynich plants keep inviting us to ask for their names.

Perhaps that is one question too early.

First ask how the picture was made.

What was copied?

What was exaggerated?

What was omitted?

What repeats?

What changes with the text?

What changes with the physical sheet?

What known manuscript traditions show us about visual transmission?

Then ask the name.

Because the real achievement will not be recognising one familiar leaf.

It will be recovering the rules that made an entire forgotten visual language useful to the people who once knew how to read 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

What the Pictures Can and Cannot Tell Us — return from one plant family to the general visual-evidence discipline.
The Vessels and Plant Fragments — test whether whole-organism and component-scale representations form a real internal bridge.
The Carrara Herbal — use a known historical comparator for plant-image/text integration without treating it as Voynich provenance.

Deep bridge: Why a Few Marks Become a Face is the perception control: familiar morphology can trigger recognition before the full feature set earns identification.

Voynich | Everything eduKate Knows and Tested | The Herbal Pages