Masson, Sloane and Roccabonella as Voynich Image-Transmission Controls

An image can survive the book that first made it useful.

That may be one of the most important lessons in the entire Voynich comparator programme.

A plant image can be copied into a new manuscript.

The old prose can disappear.

The language can change.

The order can change.

The commissioner can change.

The new book can perform a different job.

And yet the inherited image can remain recognisable.

This article examines two separate documented corridors.

Manfredus / BnF Latin 6823 → Masson 116 → Sloane MS 4016.
Carrara Herbal → selected visual adaptation in the Roccabonella Herbal.

They are not one manuscript family.

They are two independent controls demonstrating the same higher-order phenomenon:

visual information can be re-hosted while textual, linguistic, organisational and functional context changes around it.

That single fact makes many Voynich source claims harder.

And much more interesting.

Quick Read

Masson 116 is catalogued by the BnF as an incomplete copy of the Tractatus de herbis of Manfredus de Monte Imperiali in BnF Latin 6823.
The Beaux-Arts/Biblissima description says most of Masson’s plant iconography was copied from Latin 6823, while some animals and figurative scenes draw on material that follows the Manfredus plant text.
Masson 116 survives largely as images with short labels rather than the full prose medical source field preserved in Latin 6823.
The British Library catalogues Sloane MS 4016, c.1440 and associated with Lombardy, as a copy of Masson 116.
Sloane therefore lets us study what a later known copy preserves after the first host has already lost or omitted major textual context.
The Roccabonella Herbal is a different fifteenth-century manuscript, now Venice, Biblioteca Nazionale Marciana, Lat. VI, 59 (=2548), written by physician Nicolò Roccabonella and illustrated by Andrea Amadio.
The Marciana/Veneto catalogue records hundreds of entries with plant names in several languages, including Greek, Arabic, German and Croatian, plus medical properties and sources.
Sarah R. Kyle’s research reports that Amadio copied more than fifty images from the earlier Carrara Herbal.
Roccabonella therefore shows selected images from one medical book entering a new text, new order and new information architecture.
The two corridors demonstrate image survival without total book survival.
They do not prove that Voynich borrowed from either corridor.

Comparator Firewall

Masson–Sloane and Carrara–Roccabonella are two separate calibration corridors. They are not being merged into one genealogy, and neither is being extended into Voynich without evidence.

The job of the article is to learn what historical re-hosting looks like when we actually know it happened.

Corridor One Begins With an Integrated Medical Source Field

BnF Latin 6823 contains the Manfredus Tractatus de herbis et plantis inside a much larger medical environment.

Plant imagery and prose coexist with animal materia medica, antidotary material, doses, substitutions and synonyms.

The important point is not simply that Latin 6823 contains pictures.

The pictures occupy an interpretable source field.

We know what kind of book world surrounds them.

That gives us a baseline state.

Masson 116 Keeps Much of the Image System and Loses Much of the Explanatory System

Masson 116 is extraordinary because it makes selective survival visible.

The BnF describes it as an incomplete copy of the Manfredus Tractatus de herbis. Biblissima/Beaux-Arts describes its majority plant iconography as copied from BnF Latin 6823.

Yet the surviving manuscript is essentially a natural-history image collection with short Latin legends and no completed running prose.

The page still remembers an information system that the text no longer fully supplies.

This creates an important state transition:

integrated text–image medical source → image-led host with labels

That transformation teaches us that image identity and book function can separate.

A copied medical image does not carry its original prose automatically.

It does not carry its original ordering automatically.

It does not carry its original use automatically.

Sloane 4016 Copies a Book That Was Already a Transformation

The British Library’s record for Sloane MS 4016 gives us the next stage.

It is an Italian herbal of about 1440, associated with Lombardy and classified as a copy of Masson 116.

Its folios contain full-page coloured plant miniatures with captions, often accompanied by animals or people.

Notice what that means.

Sloane is not simply copying “the original medical book”.

It is copying a transformed visual host.

source field → selective image host → later copy of the selective image host

Transmission therefore compounds transformation.

A later copy can inherit the decisions of an earlier re-hosting event.

A Descendant Can Preserve What Its Immediate Parent Preserved

This sounds obvious, but it has major consequences.

If Masson omitted prose, Sloane does not need access to the earlier prose source in order to reproduce Masson’s visual lineage.

If Masson changed proportions, Sloane may inherit those changes.

If Masson selected particular images, Sloane’s surviving repertoire may reflect Masson’s selection rather than the original source field.

This creates a genealogical warning:

A late manuscript can preserve an old image tradition without preserving direct access to the old text that first accompanied it.

Corridor Two: Carrara Enters a New Book Rather Than Producing a Clone

The Roccabonella Herbal gives us a second, independent kind of re-hosting.

The manuscript is now Venice, Biblioteca Nazionale Marciana, Lat. VI, 59 (=2548).

The Marciana identifies it as the Liber de simplicibus of physician Nicolò Roccabonella, with Andrea Amadio as illustrator.

The catalogue records hundreds of folios devoted to medicinal simples, with names in multiple languages and information about their medical properties and sources.

This is not Carrara Herbal Part Two.

It is its own large medical book.

Yet Sarah Kyle’s research reports that Andrea Amadio copied more than fifty images from the Carrara Herbal.

Selected visual information crossed into a new host.

The Carrara Image Does Not Bring the Whole Carrara Book With It

This is the key.

The Carrara Herbal is a late-fourteenth-century Paduan vernacular medical manuscript with its own Serapion-derived textual architecture, patronage and history.

Roccabonella is a mid-fifteenth-century medical compilation created by a physician and an illustrator in another host architecture.

Its botanical names can be multilingual.

Its chapters have their own structure.

Its source citations and medicinal information belong to Roccabonella’s compilation project.

Yet selected Carrara images can still be reused.

Carrara visual programme → selected copied images → Roccabonella text and classification environment

The image survives.

The information architecture around it changes.

Two Corridors, Two Different Kinds of Survival

Now compare the two controls.

Manfredus → Masson → Sloane

integrated source field;
large-scale selective visual re-hosting;
subsequent known copy of that re-hosted visual tradition.

Carrara → Roccabonella

existing medical image programme;
selected images copied by a later illustrator;
new text, new compiler, new ordering environment and multilingual naming system.

The first demonstrates a visual corpus surviving after major textual context disappears.

The second demonstrates selected visual units migrating into a richly textual new book.

Together they prove something much broader than either individual genealogy.

image ancestry and text ancestry can be different.

This Breaks a Common Voynich Assumption

A Voynich researcher finds a plant that resembles a known herbal image.

The next leap is often:

therefore the Voynich text beside it is related to the text beside the source image.

Historical re-hosting shows why that does not follow.

The image can travel without its prose.

The image can enter a new language.

The image can enter a new classification system.

The image can enter a later book whose function has changed.

Therefore:

visual descent does not automatically imply textual descent.

And Textual Difference Does Not Automatically Refute Visual Descent

The reverse error is also common.

If two manuscripts use different language, order or prose, researchers may dismiss a real visual relationship.

Roccabonella shows why this is unsafe.

A new compiler can borrow selected illustrations while building a different information system around them.

Therefore a strong comparison should keep several ancestry channels separate:

image ancestry;
text ancestry;
language ancestry;
classification ancestry;
book-design ancestry;
ownership history.

They can converge.

They can also split.

What Does an Image Preserve When It Travels?

This is where known relationships become empirical controls.

We can measure:

root topology;
leaf arrangement;
flower position;
overall silhouette;
relative proportions;
presence of animals or figures;
page placement;
sequence;
colour;
labels;
rare errors or peculiarities.

Some features are more stable than others.

That stability profile is exactly what Voynich source comparisons need.

Without it, researchers often compare the wrong features.

Shared Error Is More Valuable Than Shared Beauty

Two artists can independently draw a round leaf.

Two manuscripts sharing the same highly unusual, biologically unnecessary distortion are more interesting.

This is the logic of genealogical evidence.

Generic resemblance has many causes.

Rare shared structure has fewer.

A source claim should therefore weight rare inherited features more strongly than broad medieval style.

The most useful copying clue may be the strange mistake both manuscripts share.

Order Can Survive, Drift or Be Rebuilt

Images do not travel only as isolated units.

Sequence can also be inherited.

Or lost.

Or reconstructed by a new classification system.

This matters because a proposed Voynich source relation becomes much stronger if several image correspondences appear in a related order.

Yet Roccabonella reminds us that a new compiler may reorganise borrowed images.

Therefore order is evidence whose expected stability depends on the kind of re-hosting being proposed.

Re-Hosting Is a Better Model Than “Copying” for Some Historical Transfers

The word copy can imply too much sameness.

Re-hosting is sometimes more accurate.

A component is moved into a new host.

The host supplies new context.

Think of an image inserted into a new textbook.

The pixels may be the same.

The lesson can be different.

Medieval manuscript transmission can perform a slower, hand-made version of the same operation.

This Creates a Stronger Voynich Question

Instead of asking:

Which manuscript is Voynich copied from?

ask:

Which information layers, if any, show evidence of ancestry—and did those layers necessarily travel together?

A Voynich plant may have visual ancestry in one tradition.

Its text may have another ancestry.

The page design may be local.

The manuscript may combine several source fields.

Those are hypotheses.

Known re-hosting tells us they are historically possible without telling us they happened here.

What These Image-Transmission Corridors Can Legitimately Teach Voynich Research

Medical images can survive after their original prose context changes or disappears.
A later manuscript can copy a transformed intermediary rather than the earliest source.
Selected images can migrate into a new medical book with different text and classification.
Visual ancestry and textual ancestry are separable variables.
Language difference does not automatically refute visual inheritance.
Visual similarity does not automatically transfer meaning from source text to destination text.
Known copy chains let us measure which visual features survive real manuscript transmission.
Rare shared distortions and repeated correspondence are stronger genealogical evidence than generic style.
Order may be inherited, disrupted or deliberately rebuilt depending on the host.
A useful Voynich source theory should specify which information layer it claims was transmitted.

What They Do Not Prove

that Voynich belongs to the Manfredus–Masson–Sloane chain;
that Voynich belongs to the Carrara–Roccabonella visual corridor;
that a visually related image carries the same text or meaning;
that Roccabonella and Masson belong to one genealogy;
that every Voynich distortion is the result of copying;
that a missing textual parallel can always be excused as re-hosting;
that northern Italian image transmission proves northern Italian Voynich production;
that visual lineage alone identifies the language of Voynichese.

A Re-Hosting Test for Voynich

Choose documented source–copy pairs and quantify their transformation before testing Voynich.
Separate image, text, order and layout features.
Measure which features are stable across Manfredus–Masson–Sloane.
Measure selected image adaptation across Carrara–Roccabonella.
Define a transformation budget from those real examples.
Apply the same budget to proposed Voynich correspondences.
Require several independent image matches, not one showcase pair.
Search for rare shared distortions and ordering evidence.
Do not infer textual meaning from visual ancestry unless text ancestry is demonstrated separately.
Reject a source claim if it requires substantially more freedom than real known re-hosting corridors.

Student Lens: The Same Picture Can Teach a Different Lesson

Primary: place the same photograph in a science book, travel guide and advertisement. The image stays the same while its function changes.

Lower Secondary: compare two documents that reuse the same diagram with different captions. Ask what information moved and what did not.

Upper Secondary: model copying as selective transmission: source features have different probabilities of surviving into a descendant.

JC and adult readers: treat manuscript lineage as a multi-channel inheritance problem in which image, text, language, ordering and book function may follow partially independent paths.

Research Sources

Continue the Image-Transmission Route

The Final Idea

The history of a medical image is not necessarily the history of the book around it.

An image can leave home.

It can lose its prose.

It can be copied from an already transformed intermediary.

It can enter a new language and a new classification.

It can remain recognisable after the first intellectual architecture around it has disappeared.

If Voynich inherited visual material from an older medical tradition, the correct target is not a perfect twin. It is a historically plausible transformation path whose surviving invariants are stronger than coincidence and whose freedoms are no greater than real manuscript transmission allows.

Voynich Comparator | When Medical Images Outlived Their First Books: Masson, Sloane and Roccabonella