Morgan M.873 as a Voynich Comparator

Morgan M.873 looks, at first glance, like the kind of book that can make a Voynich researcher dangerously confident.

There are plants. There are labels. There are animals, minerals, vessels, human activities and substances drawn for recognition. There is a late-medieval Italian medical world sitting visibly on parchment.

That makes it useful.

It also makes it dangerous.

Comparator is not source. Resemblance is not descent. A similar-looking plant does not establish copying, provenance, language, author or purpose.

Morgan M.873 belongs in the eduKateSG Voynich research library precisely because its historical position is much better constrained than the Voynich Manuscript. It gives us a control object: a real medieval medical image system whose date, broad region, manuscript family and later reception can be studied independently.

Quick Read

The Morgan Library & Museum catalogues M.873 as the Compendium Salernitanum.
The Morgan dates it to 1350–1375 and places its production in northern Italy, possibly Venice.
The surviving volume is strongly image-led and contains hundreds of representations of medicinal plants and other therapeutic materials.
Its visual field extends beyond plants to animals, minerals, human activities and processed substances.
Many depicted objects are labelled, so image and name can be compared directly.
The manuscript belongs to the wider late-medieval Tractatus de herbis / Salernitan materia-medica tradition rather than being an isolated botanical sketchbook.
Its relationship to better-textualised medical traditions shows that images could travel partly independently of explanatory prose.
For Voynich research, M.873 is valuable as a control for layout, visual transmission, label behaviour, materia-medica scope and the separation of image from text.
It does not establish that the Voynich Manuscript was copied from M.873, made in Venice, written in Latin, or produced for the same users.

The First Thing to Fix: What M.873 Actually Is

The Morgan’s digital manuscript catalogue identifies the object as Compendium Salernitanum, M.873, produced in Italy, possibly Venice, between about 1350 and 1375.

That description already tells us several things the Voynich Manuscript does not tell us about itself.

the manuscript sits inside a recognisable medical-literary tradition;
its approximate date is supported by codicological and art-historical work;
its images can be compared with related manuscripts;
many objects can be associated with conventional medieval materia-medica names;
there is a surviving network of predecessor, parallel and successor witnesses against which copying and transformation can be tested.

This is the key methodological advantage.

With M.873, we can observe what an actual manuscript family looks like when it survives well enough to be reconstructed.

It Is a Medical Material World, Not a Modern Botany Book

Modern readers tend to see a plant picture and ask, “What species is that?” Medieval medical books often asked a wider question.

What useful substance is this, by what name can it be recognised, and how does it enter a therapeutic system?

That system can include plants, roots, resins, seeds, fruits, stones, minerals, animal products and manufactured materials. M.873 therefore contains a broader inventory than a modern herbarium.

This matters for Voynich interpretation. A page dominated by a large plant-like figure should not automatically be classified as “botany” in the modern disciplinary sense. In medieval materia medica, the plant may be one node in a network joining identity, trade, preparation, substitution, dosage and treatment.

M.873 gives us a documented example of that wider world.

The Pictures Are Not Isolated Acts of Observation

One of the easiest mistakes in historical image analysis is to assume that every drawing represents a fresh encounter between an artist and a living specimen.

Medieval medical images could be copied from earlier manuscripts, corrected from observation, simplified, embellished, misunderstood, relabelled or moved into a new page design. One image can therefore contain several historical layers at once:

an inherited iconographic skeleton;
a workshop’s drawing convention;
a local artistic style;
a remembered or observed botanical feature;
a copied error;
a practical identifying feature;
a symbolic or narrative addition.

M.873 is useful because its place inside the illustrated medical tradition gives scholars something to compare those layers against.

For Voynich work, this means that “the root looks strange” or “the leaf resembles species X” is not enough. We need to know whether the feature is stable across related figures, whether it is diagnostic, whether it belongs to an inherited drawing convention, and whether the same visual grammar occurs elsewhere.

A Known Family Teaches Us What Real Descent Looks Like

This is perhaps M.873’s most important contribution to Voynich methodology.

When manuscripts genuinely belong to the same transmission family, the evidence does not usually depend on one attractive resemblance. It accumulates across many dimensions.

shared sequences of entries;
recurring iconographic poses;
the same unusual objects;
stable errors or distortions;
parallel labels or synonyms;
similar grouping rules;
corresponding omissions;
recognisable transformations from one copy to another.

The more independent correspondences survive, the stronger the stemmatic case becomes.

This gives us a standard for Voynich claims.

If a proposed Voynich source relationship cannot produce a network of repeated, ordered, independently testable correspondences, it has not yet reached the evidential level of an ordinary manuscript-family argument.

Image and Text Can Separate

The wider manuscript family around M.873, Manfredus/Latin 6823, Masson 116 and Sloane MS 4016 demonstrates something that is easy to underestimate: medieval books were modular.

A later maker did not have to copy an earlier book as one indivisible package.

An image could travel without its full prose explanation.
A label could be retained while synonyms changed.
Several images could be regrouped onto one page.
A textual section could be omitted.
Page layout could be redesigned while image identity remained recognisable.
A copied image could acquire new decorative or social detail.

That is why Masson 116 can survive as a largely text-missing image corpus and still feed later visual production, and why Sloane 4016 can reorganise inherited imagery into a different page system.

M.873 helps us see that image transmission and textual transmission can have different survival paths.

Why This Matters for the Voynich Herbal Pages

The Voynich Manuscript contains many pages that modern cataloguers describe as herbal because a large plant-like figure dominates the page.

M.873 supplies several useful controls.

Control 1: How does a real medicinal image inventory distribute object types?

Do plants remain isolated, or do animals, minerals, vessels and human actions appear in the same information world?

Control 2: What does a known label system look like?

When the labels are readable, do they name the pictured object, a synonym, a substance, a preparation, a location or something else? This tells us how much meaning one should expect from label placement alone.

Control 3: What survives copying?

Which features remain stable when an image moves into a later manuscript? Stable features are better candidates for functional identity than details that change with workshop style.

Control 4: What happens to order?

Medical image traditions can preserve or transform entry order. If a proposed Voynich relationship is real, sequence-level correspondences should be testable, not just single-page resemblance.

Control 5: How independent are image and prose?

M.873 belongs to a world in which illustration can remain useful even when explanatory text is reduced, moved or absent. That makes it a better comparator for information architecture than a fully prose-dominated medical codex alone.

The Critical Difference: M.873 Is Legible to Its Historical Neighbours

The strongest contrast with Voynich is not that one manuscript has plants and the other has stranger plants.

The strongest contrast is that M.873 sits in a decipherable ecosystem.

Its manuscript family can be compared.
Its medical vocabulary belongs to known learned traditions.
Its imagery can be matched against related witnesses.
Its geographic and chronological placement can be debated within a bounded historical frame.
Its later ownership and annotations can be distinguished from its original production.

Voynich lacks that degree of external anchoring.

That makes the correct research move obvious: use M.873 to learn how a documented visual-medical system behaves, then ask whether Voynich independently shows the same mechanisms.

Do not reverse the logic and declare that because the mechanisms are possible, Voynich must belong to this family.

A Stronger Test Than “These Two Plants Look Alike”

A credible comparison between M.873 and a Voynich subset should pre-register features before looking for a match.

root architecture;
leaf attachment pattern;
flower or fruit placement;
orientation of the whole specimen;
relative scale of plant parts;
presence and placement of labels;
page-level grouping;
adjacent entry order;
recurrence of the same motif elsewhere;
frequency of decorative versus identifying detail.

Then compare the Voynich image not only with the proposed M.873 counterpart but with a large set of unrelated medieval images.

If the claimed similarity remains exceptional after those controls, it becomes interesting.

If it disappears once common medieval drawing conventions are included, it was never strong evidence.

M.873 and the Problem of Missing Text

A visual medical manuscript can preserve useful knowledge without preserving the full explanatory prose that once accompanied, preceded or paralleled the image tradition.

But absence must not be filled by imagination.

Prepared or inherited visual structure can demonstrate that an information layer existed. It cannot recover the exact wording of a missing layer.

This is one of the most transferable lessons from the Masson–M.873–Sloane comparison family. A book may tell us that pictures, labels and ordering were functional without telling us the exact sentences a lost or omitted text would have contained.

The same discipline must govern Voynich. If an image arrangement suggests indexing, classification or recipe structure, that is a hypothesis about function. It is not permission to invent plaintext.

What M.873 Can Legitimately Teach Voynich Research

Medieval medical image systems can be much broader than modern botanical books.
Images can function as identifiers, memory aids, prestige objects and transferable modules simultaneously.
One manuscript can preserve an image tradition even when prose transmission changes.
Plant pictures may be conventionalised through copying rather than freshly observed.
Labels can be structurally useful without being long prose.
A real manuscript family should generate repeated correspondences across many entries, not a handful of visual coincidences.
Entry order and omission patterns are evidentially valuable.
Later annotations and ownership layers must be separated from original production.
Visual resemblance is most useful when it produces a falsifiable prediction elsewhere in the corpus.

What M.873 Does Not Establish

that Voynich is a copy of M.873;
that Voynich was produced in Venice;
that Voynich is a Salernitan medical text;
that Voynich plant drawings can be identified by visual resemblance alone;
that the Voynich script encodes Latin plant names;
that the Voynich sections correspond to the same textual genres as the Compendium Salernitanum;
that a text-light medical book and an unreadable text-rich book have the same function;
that chronological proximity proves contact.

The Better Research Question

Instead of asking:

Is the Voynich Manuscript another M.873?

Ask:

Which observable mechanisms of visual-medical information transfer in M.873 also occur in Voynich, and which do not?

That question can fail cleanly.

It can discover difference.

And difference is often more valuable than resemblance.

A Practical Comparator Protocol

Freeze the Voynich sample before opening the comparator.
Record only observable image features first.
Separate page layout, object form, labels and sequence.
Compare against several manuscripts, not only M.873.
Use known manuscript-family correspondences to calibrate what genuine descent looks like.
Score negative matches and missing correspondences, not only successes.
Do not convert a visual match into a language or provenance claim.
Require a proposed mechanism to predict unseen pages.

This is consistent with the wider eduKateSG Voynich runtime: blind structure first, interpretation after the scoring rule is frozen.

Primary to Secondary: What Students Can Learn From This Manuscript

Primary: two drawings can look similar without being copies. Students can compare leaf shape, root shape and labels and learn the difference between observation and inference.

Lower Secondary: students can track one motif across several manuscripts and mark which features remain stable and which change.

Upper Secondary: students can design a blind comparison test in which one group selects features and another scores candidate matches.

JC and adult readers: M.873 becomes a case study in transmission theory, stemmatics, information modularity, iconographic inheritance and the statistical problem of multiple comparisons.

Research and Primary Sources

The Final Idea

Morgan M.873 is valuable because it is not mysterious in the same way Voynich is mysterious.

It shows us what becomes visible when enough of a medieval information system survives around an object.

We can see the family.

We can see the transfer.

We can see images moving partly independently from prose.

We can see that medicinal knowledge was a world of substances, names, pictures, trade and practice rather than a modern botany textbook.

And because we can see all of that, M.873 gives Voynich research something more useful than a possible answer.

It gives us a standard for what evidence should look like when a real historical relationship exists.

Voynich Comparator | Morgan M.873: The Compendium Salernitanum and the Medical Image Inventory