Voynich and Padua | Hand × Bifolio × Section: Reconstructing the Production Matrix

The Voynich Manuscript gives us several ways to divide itself.

By pictures.

By scribal hand.

By Currier textual population.

By bifolio.

By quire.

By current section.

The mistake is to assume these partitions are the same.

The production problem is not “which section came first?” It is “which physical, scribal, textual and visual partitions actually align, and which cross one another?”

Quick Read

Lisa Fagin Davis identified five scribal hands with high confidence.
Prescott Currier identified major statistical textual populations traditionally called A and B, while explicitly warning that “language” was only a convenient label.
In the herbal material, Currier A/B behaviour tracks whole bifolios rather than arbitrary page boundaries.
The current visual “sections” do not map perfectly onto either hand or textual population.
Some hands cross conventional section boundaries.
Some sections contain more than one textual population or hand.
The bifolio is especially important because it is a physically inseparable production unit unless cut.
Therefore the best public production model is multidimensional: hand × bifolio × text state × visual province × current order.
This matrix can reveal whether Voynich is one coordinated project, several modules, or a composite object whose present sectioning obscures production history.

One Manuscript, Several Partitions

Modern readers first see the illustrations.

That creates the familiar sections:

herbal;
astronomical/cosmological;
zodiac;
biological/balneological;
pharmaceutical;
starred or recipe-like text.

But pictures are only one axis.

The scribes divide the manuscript differently.

Currier’s statistics divide it differently again.

The bifolios impose a physical partition underneath both.

If these layers disagree, the disagreement is information.

The Five Hands Are a Palaeographic Layer

Davis’s five-scribe model gives the manuscript a much clearer collaborative architecture.

The hands are not evenly distributed by modern subject label.

One hand can appear across several conventional content regions.

This matters because it weakens any simple model in which one specialist writes one subject.

A scribe may copy several modules.

A workshop may allocate work by physical batch rather than subject.

Or one trained operator may genuinely work across several domains.

Currier A and B Are a Textual Layer

Currier discovered that substantial portions of the manuscript differ statistically in repeatable ways.

He called the two major populations Language A and Language B while cautioning that they need not be separate languages.

Possible explanations remain open:

different languages;
dialects;
subject matter;
scribal habits;
encoding variants;
production states.

The labels A and B are therefore observations about statistical state, not recovered semantics.

The Herbal Bifolios Are the Crucial Public Control

One of the strongest publicly documented structural facts is that the A/B distinction in the herbal material follows bifolios.

Whole bifolios are in one textual state rather than flipping between A and B across the two leaves.

This tells us that whatever creates the statistical difference is tied to production at a physical-unit scale.

The text state knows about the bifolio.

That is much stronger than a page-by-page thematic fluctuation.

Why the Bifolio Matters More Than Current Adjacency

Two leaves on one bifolio were created from the same physical sheet.

They remain linked even if the sheet is restacked elsewhere in the quire.

Current facing pages, by contrast, can change after rebinding.

Therefore the evidence hierarchy should be:

bifolio relationship > current neighbouring-page relationship.

This is why every textual and visual classification should be projected onto the physical collation.

Hand and Currier State Are Related but Not Identical Concepts

Currier’s original two principal hands correlated strongly with his A and B populations.

Davis’s later five-hand palaeography gives a higher-resolution scribal map.

The right question now is not whether “hand equals language”.

It is:

which textual properties are stable across hands;
which change with hand;
which change with visual section;
which are locked to bifolio;
which survive all of those boundaries.

That is the production matrix.

A Stable Feature Across Hands Is More Likely System-Level

Suppose a boundary rule appears in all five hands.

That suggests it belongs to the shared script or encoding system.

Suppose a word family belongs only to one hand.

That raises a scribal-style hypothesis.

Suppose a visual component occurs across several hands but only in one functional page class.

That raises a content-function hypothesis.

Crossing axes lets us assign claims to the right layer.

A Feature Locked to a Bifolio May Be Production-State Evidence

If a textual feature remains stable across both sides of a bifolio but changes at the next physical sheet, the explanation may sit at the production level.

Possible causes include:

different source unit;
different scribe;
different encoding setting;
different production session;
different intended module.

Again, physical alignment narrows explanation without revealing semantics.

Modern Sections Can Cut Across Production Units

The familiar Voynich sections are based primarily on illustration type.

But a visual transition does not necessarily equal a scribal transition.

A hand can continue across one.

A textual state can cross another.

A bifolio may contain images that modern viewers place in the same broad section but belong to a distinct production batch.

This means “section” should be treated as a descriptive layer, not a canonical production unit.

The Current Order Adds Yet Another Distortion

The original-order problem makes the matrix even more necessary.

If bifolios were restacked before numbering, current section adjacency can join units that were not originally adjacent.

Therefore any analysis that smooths features across current page sequence risks mixing production states.

The matrix protects against that by anchoring every observation to the physical sheet first.

The Matrix Has Five Public Axes

Scribal hand — Davis 1–5.
Bifolio — physical production unit.
Text state — Currier A/B or another explicitly described statistical classification.
Visual province — herbal, circular, human/tube, pharmaceutical, starred and other descriptive classes.
Current location — present quire and folio order.

No private score or semantic decoding is required to build this public matrix.

What Alignment Would Mean

Perfect Hand × Text-State Alignment

Scribal implementation could explain much of the textual variation.

Text State Crosses Hands

The state is probably larger than individual scribal habit.

Visual Province Crosses Hands and Text States

The visual subject may be compiled independently of script implementation.

Everything Locks to Bifolio

Production batching becomes the dominant explanation.

What Misalignment Would Mean

Misalignment is not noise.

It can expose the workflow.

If one hand writes two textual states, that scribe knew both states.

If two hands write the same state, the state is shared.

If one visual grammar crosses several states, the image system may sit above text-state variation.

The intersections are more informative than any one partition alone.

What This Does to the “Integrated Manual” Hypothesis

A truly integrated manual should preserve some invariants across modules.

Those invariants might be:

boundary grammar;
label morphology;
visual component grammar;
page planning;
shared scribal conventions.

If no invariants survive across modules, the composite-book hypothesis grows stronger.

If several survive despite hand and topic changes, a coordinated production system becomes more plausible.

What This Does to the Padua Hypothesis

Padua becomes relevant only after the production matrix stabilises.

Then we can ask:

Does the collaboration scale resemble Paduan manuscript production?
Do hand clusters match Paduan scribal controls?
Do bifolio practices match Paduan codices?
Do text-state modules resemble Paduan technical genres?
Do the same results beat Bologna, Venice, Lombardy and Central Europe?

Without that comparison, the matrix is production evidence but not provenance.

Public Production-Matrix Test

Assign each folio to Davis hand where available.
Assign each folio/bifolio to Currier text state where available.
Anchor both to physical bifolio.
Add neutral visual-province labels.
Add current quire position.
Measure alignment and crossing between axes.
Repeat key text features by hand and bifolio.
Only after stable patterns emerge compare historical production environments.

What the Production Matrix Can Legitimately Support

the manuscript contains several scribal hands;
it contains repeatable textual populations;
in important regions, textual state aligns with bifolio;
modern visual sections are not sufficient production categories;
cross-axis analysis can separate scribal, physical and functional effects.

What It Cannot Support

that Currier A/B are two languages;
that each scribe was a separate author;
that one visual section equals one production module;
that production alignment reveals semantic topics;
that any published public matrix exposes or validates private experimental results.

Research Sources

Continue the Series

The Final Idea

The Voynich Manuscript is not divided one way.

It is divided several ways at once.

The production system will become visible when we stop asking which partition is “the real one” and start asking which partitions lock together, which cross one another, and what kind of workflow could have produced that matrix.