Voynich Production Order: Drawing, Writing, Paint and Assembly

A manuscript has a history before it has a meaning.

Someone prepares parchment.

Someone draws.

Someone writes.

Someone adds colour.

Sheets are stacked.

Gatherings are numbered.

The book is sewn.

Pages are removed.

Later readers add their own marks.

By the time we meet the Voynich Manuscript in Yale’s Beinecke Library, all of those events have collapsed into one object.

But they did not happen at once.

The parchment preserves overlaps.

Ink avoids drawings.

Words fit into spaces between plant branches.

Paint crosses outlines and, in places, appears to sit over writing.

Folio and quire numbers occupy their own historical layers.

Missing leaves and binding evidence tell us that assembly changed after some earlier stages were already complete.

This allows one of the most useful kinds of Voynich research:

reconstruct the order of physical operations before trying to reconstruct the order of ideas.

That distinction matters because production sequence can constrain interpretation without translating a single glyph.

If drawings came first, the text was not simply illustrated afterward in the ordinary modern sense.

If text was fitted around pre-existing outlines, page geometry was part of the writing task.

If colour was added later, colour cannot automatically be treated as information generated at the same moment as the outlines and text.

If the present order was disturbed before later numbering and binding stages, current adjacency cannot automatically be treated as original conceptual sequence.

The book begins to separate into historical layers.

That is progress.

Quick Read

One-sentence answer: the strongest production evidence indicates that, on individual bifolia and foldouts, illustration outlines were generally made first, the Voynich text was then written around and between those outlines, and colour was applied afterward; beyond that local sequence, proposed timelines involving rearrangement, numbering, partial painting, disassembly and rebinding are valuable but increasingly tentative reconstructions rather than settled history.

The Voynich Manuscript was written on prepared calf parchment dated by radiocarbon analysis to 1404–1438 at 95% probability.
The parchment does not show evidence that an earlier text was erased; it is not a palimpsest.
On many pages, text visibly avoids illustration outlines.
In some places, words fit tightly between branches or other drawn structures.
This strongly supports outline-before-text production for many illustrated pages.
It does not prove that every drawing on an entire bifolium was completed before any text was written.
Page-by-page and bifolium-wide working sequences remain difficult to distinguish.
Paint frequently sits over outlines and, in some locations, apparently over text, supporting paint-after-ink in those loci.
The exact timing of painting across the whole codex remains one of the main uncertainties.
Some researchers have proposed multiple painting stages or painters; these remain hypotheses rather than settled fact.
Currier and later observations make bifolium-by-bifolium production a strong working reconstruction.
The present order of folios is very likely not the originally planned order.
Quire and folio numbering belong to later organisational layers and were added by different hands.
A detailed modern reconstruction proposes rearrangement before numbering, painting before the present binding, and loss of several bifolia before the current sewing; parts of that sequence remain tentative.
Colour should therefore be treated more cautiously than outline and text when used for plant identification or semantic inference.

The safest way to read the production evidence is as a confidence ladder.

Strong: outline and text are separate operations, with outline generally preceding text on illustrated pages.
Strong: colour was applied after ink in many observable places.
Strong working reconstruction: bifolia were meaningful production units.
High-confidence codicological inference: current page order differs from the originally planned order.
More tentative: the exact sequence of rearrangement, numbering, partial painting, disassembly, missing bifolia and rebinding.

Keeping those levels separate is more useful than pretending the entire timeline has equal certainty.

Production Evidence Is Different From Content Evidence

If a plant has five leaves, that may or may not tell us what species the artist intended.

If a line of text stops before a leaf outline and resumes after it, the evidentiary question is different.

We are not asking what the leaf means.

We are asking which physical mark was already present when the scribe wrote the line.

That is a stratigraphic question.

Archaeologists use stratigraphy when one deposit lies above another.

Manuscript researchers use the same logic at microscopic scale.

If stroke A crosses stroke B, and the material relationship is clear, the crossing can establish relative order.

Relative order is often much easier to establish than absolute date.

We may not know whether two operations were separated by ten minutes, ten days or ten years.

But we may know which came first.

This is exactly the kind of bounded knowledge Voynich research should value.

Stage 0: The Parchment Was Prepared Before the Mystery Began

The manuscript’s writing surface is calf parchment.

Four samples taken from different bifolia produced mutually consistent radiocarbon results, giving a combined 1404–1438 parchment range at 95% probability.

The important distinction remains:

radiocarbon dates the parchment, not the exact moment the Voynich script was written.

Even so, unused parchment was valuable, and long storage before use would require its own historical explanation.

The parchment also carries production information.

Some sheets contain holes or stitched defects inherited from parchment making.

The surfaces were prepared carefully enough that hair and flesh sides can be difficult to distinguish in many places.

Microscopic and multispectral examination has not revealed erased earlier writing, so the codex is not a reused palimpsest whose present content sits over another book.

That matters.

The production sequence begins on parchment prepared for this writing event, not on a scraped-away previous text that must be reconstructed underneath.

Stage 1: Illustration Outlines Generally Came Before Text

This is one of the strongest local production conclusions.

On illustrated pages, the text repeatedly behaves as if the drawing already exists.

Lines avoid stems.

Words occupy narrow spaces between branches.

Text bends around visual structures.

The simplest physical explanation is:

the outline was already on the parchment when the text was placed.

This does not tell us whether the artist and scribe were different people.

One person can switch tools and roles.

Several people can work from one prepared model.

The evidence establishes operation order more strongly than personnel.

That distinction keeps the production argument separate from the scribe-count debate.

Outline First Does Not Mean “The Pictures Explain the Text”

This is an important semantic firewall.

If the drawing was present first, the scribe had to respond to its geometry.

That does not prove the text describes the drawing.

A label probably has some relation to what it is placed beside.

A paragraph beside a plant may discuss the plant.

But it could also discuss:

uses;
classification;
substitution;
timing;
preparation;
a cross-reference;
something not represented literally by the picture.

Physical precedence establishes page planning.

It does not establish semantic equivalence.

Text Fitted Into Prepared Visual Space

The most intriguing cases are those in which the available visual gap fits the written word closely.

A branch leaves a narrow corridor.

A Voynich token occupies the corridor.

The fit can look almost suspiciously precise.

This has encouraged a strong interpretation:

perhaps the writer generated meaningless text simply to fill the remaining space.

That is one possible mechanism.

It is not the only one.

A prepared draft can solve the same spatial problem.

The composer may know in advance what text belongs in a region and design the drawing accordingly.

The text may use abbreviation or flexible spelling to fit available space.

The scribe may have copied from a model in which image and text were already integrated.

Therefore:

tight fit proves spatial coordination, not meaningless filler.

The fourth article in this batch owns that page-planning problem in depth.

We Do Not Know Whether All Four Bifolium Surfaces Were Drawn Before Any Were Written

The outline-before-text conclusion is local.

Scaling it up introduces uncertainty.

A standard bifolium eventually contributes four written surfaces.

How did the maker work?

Workflow A

Draw all four surfaces, then write all four.

Workflow B

Draw one page, write it, move to the next page.

Workflow C

Prepare some drawings, write some text, return later to complete other surfaces.

Current physical evidence does not force one of these for the entire codex.

This is a good example of why production evidence must remain scale-aware.

We can know A precedes B at one crossing without knowing the complete schedule of the workshop.

Stage 2: The Text Was Written in Ink

The principal writing uses iron-gall ink, with only minor compositional variation detected across the manuscript.

That does not establish one writing session.

Iron-gall ink was common across long periods.

Minor variation can arise from different batches without implying different centuries.

More importantly for production order, the text behaves as a second integrated layer after the drawn outline.

The writer respects visual space.

Lines are not ruled formally across the page.

The writing adapts to the illustration.

That tells us something about composition style.

This is not a text block prepared independently and pasted conceptually into a finished illustration.

The writing event is spatially coupled to the page.

Stage 3: Colour Came After Ink in Many Observable Places

Colour behaves differently from outline.

Paint can lie over the drawn lines.

In some places it appears to cross text.

This supports a basic local sequence:

outline → text → paint.

Again, this is relative ordering.

It does not tell us whether paint followed an hour later or after a much longer interval.

It also does not prove one painter.

Some scholars have proposed lighter and heavier painting stages or different painters.

Those proposals remain debated.

The broad conclusion is safer:

colour is a later layer than the principal ink work in many parts of the codex.

This is why colour should carry a lower default evidentiary weight than outline geometry when identifying depicted plants or objects.

Colour May Not Belong to One Production Moment

The most important uncertainty in the production sequence concerns painting.

Several possibilities remain live.

All colour was added relatively soon after the ink work.
Some colour was added early and some later.
Different painters contributed at different times.
The manuscript was partly reorganised before colouring was completed.

Paint-transfer observations have been used to reconstruct parts of this history, especially blue pigment transferred by contact to facing pages.

These observations are potentially powerful because wet or friable pigment transferring between surfaces ties painting to physical adjacency.

But the comprehensive paint-transfer map remains incomplete and some interpretations are debated.

Therefore the correct public position is:

paint can provide ordering evidence, but the full painting chronology is not settled.

The Folio Number Inside a Painted Leaf Is a Remarkable Clue

Folio 42 contains one particularly interesting crossing of administrative and artistic layers.

The folio number sits within the outline of a leaf that was later coloured green.

Microscopic inspection reportedly indicated that green pigment lies over the folio number.

If that reading is correct, at least some painting occurred after foliation.

This is exactly the kind of evidence that complicates a simple one-session painting story.

It suggests that administrative organisation and colour application may overlap in time in ways we would not expect from a perfectly linear modern production process.

One must still resist turning one locus into a universal law.

It proves a local ordering if the microscopic interpretation is correct.

It does not prove that all colour everywhere was applied after all folio numbers.

Bifolium-by-Bifolium Production Is a Strong Working Reconstruction

The codex is made from physical sheets, not isolated modern pages.

That matters because Currier-type text properties and proposed handwriting distinctions often show coherence at the level of the bifolium.

René Zandbergen therefore argues that there is every reason to believe the manuscript was basically produced bifolium by bifolium, including its special foldout sheets.

This does not mean every operation on one bifolium was completed before the next sheet was touched.

It means the physical sheet appears to preserve coherent production state strongly enough to deserve primary analytical status.

The next article in this batch owns that proposition in depth.

The important conceptual shift is simple:

the historical maker handled sheets; the modern reader experiences pages.

Production reconstruction should start with the maker’s unit.

Stage 4: Bifolia Were Gathered Into Quires

Once sheets exist, they must become a codex.

Bifolia are stacked.

Stacks become quires.

Quires are sewn.

But the Voynich Manuscript contains strong evidence that the stack we have is not simply the originally planned sequence frozen forever.

Herbal A/B bifolia appear intermixed.

Some visually related pharmaceutical sheets are split across quires.

A Quire 13 bifolium contains an integrated design whose current location does not display it as one would expect if the intended spread were preserved.

Quire 9’s unusual foldout structure adds another historical puzzle.

These observations support high confidence that current order differs from an earlier planned order.

They do not give us a complete original sequence.

This is why every semantic argument that depends on “the next page” must remain order-sensitive.

The Page Order Was Probably Disturbed Before the Later Numbering System Was Complete

Quire and folio numbers increase through the current sequence even across places where earlier ordering appears disturbed.

This suggests the reordering happened before those numbering layers fixed the sequence administratively.

The exact timing remains complex.

Quire numbers and folio numbers were added by different people.

The quire numbers are generally considered early, likely fifteenth century.

The folio numbers are less tightly dated palaeographically.

What matters for reconstruction is that the numbering systems describe an already organised book state.

They are not labels written by the original composer onto loose sheets before any ordering decisions existed.

At minimum, numbering belongs to a later organisational layer than principal drawing and text.

Missing Leaves Give Us Relative Dates

The missing folios are not one event.

Most missing material consists of absent bifolia or entire missing quires.

Folia 12 and 74 are different: their other bifolium halves survive and the leaves were cut from the bound book.

Numbering gaps show that the people who numbered the manuscript knew about leaves that are now gone.

Therefore at least much of the missing material survived long enough to be included in an earlier numbered state.

This gives us a partial chronology without recovering any missing content.

It is another good example of evidence discipline.

We can infer:

the missing leaves existed at stage X.

We cannot infer:

therefore they contained the missing transition my theory needs.

Prepared absence is not recovered content.

The Present Binding Is Old, but Not Proven Original

Conservators examined the sewing and binding evidence in detail.

The present sewing is old and may be fifteenth century.

But it has not been established as the first binding.

Other material evidence suggests an earlier binding state existed.

This matters because “old binding” and “original order” are not equivalent.

A manuscript can be rebound early in its life.

A mistaken or revised ordering can become historically old.

Age does not restore intentionality automatically.

The current binding is evidence of one important historical state.

It is not a guaranteed window onto the composer’s first sequence.

A Tentative Reconstruction Goes Further—And Must Be Labelled Tentative

René Zandbergen has synthesised the codicological clues into a more detailed tentative production history.

In simplified form, the proposal is:

Bifolia were prepared with outline drawings and text.
The planned order was disturbed while the set was still complete.
Quire and folio numbering were added in subsequent organisational stages.
The book may have been bound in an intermediate state.
Painting was incomplete or not yet applied everywhere.
The manuscript was disassembled and painting was completed or continued.
Several bifolia were lost or removed before the present binding state.
The book was rebound, with later removal of folia 12 and 74.

This is a valuable synthesis because it tries to make all observations coexist in one timeline.

It should not be published as though each step has equal evidentiary status.

The author of the reconstruction explicitly calls it tentative and notes that production reconstruction remains incomplete.

That caveat is part of the evidence.

The right public formulation is:

this is one coherent reconstruction consistent with several observed layers, not a recovered workshop diary.

If Painting Happened After Reordering, Colour Becomes Historically Dangerous

One implication of the tentative reconstruction is especially important for visual interpretation.

If some or much painting occurred after bifolia had already been reordered—and perhaps when original compositional understanding was no longer available—then colour may not preserve the original designer’s intended coding faithfully.

This would affect plant identification strongly.

A blue root is already weak evidence of botanical identity.

If blue was added by a later painter with incomplete understanding, it becomes weaker still.

But the chain contains uncertainty.

We should not jump from “paint is later than ink” to “the painter understood nothing”.

The safest conclusion is simpler:

colour is a separate production layer whose semantic authority has to be demonstrated rather than assumed.

Production Order Can Separate Author, Scribe, Artist and Painter

Modern readers often imagine one mysterious Voynich author doing everything.

Historical manuscript production need not work that way.

Roles can be distributed.

Composer.
Draft planner.
Artist.
Scribe.
Colourist.
Assembler.
Binder.

One person may perform several roles.

Several people may share one role.

Production order helps separate them conceptually.

If the text follows a drawing, the person writing may be responding to another person’s earlier work—or to their own earlier work.

If paint is clearly later and variable in quality, colourist and outline artist become separable roles even if they ultimately prove to be the same historical individual.

The role model is safer than assigning identities prematurely.

Production Order Can Test the “Meaningless Filler” Hypothesis

Tight text fitting around drawings is often cited in favour of generated filler.

The logic is understandable.

If the writer can generate arbitrary Voynich-looking strings, filling any remaining gap becomes easy.

But the same physical observation supports at least three models.

Generated filler: create strings to occupy available space.
Prepared draft: copy a preplanned image-text layout.
Flexible meaningful notation: use abbreviation, spelling variation or compact technical units to fit known content.

The production evidence alone cannot choose.

Other text properties must discriminate.

Do tight-fit words maintain normal vocabulary relationships?

Do they show unusual abbreviation?

Are they semantically predictable under a decipherment?

Does their length correlate with available space more strongly than matched controls?

Production generates the question.

It does not decide the semantic answer.

Production Order Can Test Decipherments

A proposed decipherment should be compatible with the way the object was made.

If a theory claims the text generated the drawing instructions, but the visible drawing outline clearly precedes the text, the theory needs an earlier invisible draft or another mechanism.

If a theory relies heavily on colour as the primary semantic key, it must explain the later and possibly multi-stage paint layer.

If it treats current page order as a necessary computational pipeline, it must survive order uncertainty and apparent rearrangement.

If it requires missing leaves to contain instructions that make the sequence work, it has used absence as rescue.

A real solution should become more plausible when confronted with production history.

It should not need production history to be ignored.

What Survives the Production-Order Work

The parchment was prepared before the principal drawing and writing operations and is not a reused palimpsest.
Illustration outlines generally preceded text on illustrated pages.
Text was spatially fitted around and between existing outlines.
This proves coordination with page geometry, not meaningless filler.
Paint was added after principal ink work in many observable locations.
The exact painting chronology remains incompletely reconstructed.
Bifolia are strong production units and should be analysed as such.
The present page order is very likely not the originally planned order.
Quire and folio numbering are later organisational layers and were written by different people.
Much missing material survived long enough to be known to earlier numberers.
The present binding is old but not proven to be the first binding.
A detailed rearrangement–numbering–painting–rebinding timeline exists as a useful tentative synthesis, not settled fact.
Production sequence places strong constraints on colour-based, order-based and image-first decipherment theories.

What Does Not Survive as Established Knowledge

One person definitely drew, wrote and painted the entire manuscript.
Every outline in the manuscript was completed before any text anywhere was written.
All four surfaces of each bifolium were drawn before any of its text was added.
All colour was applied in one session.
The painter was definitely different from the original composer.
The painter definitely did not understand the manuscript.
Tight word fitting proves meaningless generated filler.
The present binding preserves the original intended order.
The complete original order has been reconstructed.
The detailed modern production timeline is established fact at every step.
Missing folios can be assigned whatever content a theory requires.

The relative layers survive.

The complete workshop biography does not.

A Better Production-Sequence Analysis

Record physical overlap before interpreting content.
Separate local order from manuscript-wide order.
Analyse the bifolium rather than only isolated pages.
Map outline, text, paint, numbering and later writing as separate layers.
Distinguish direct crossing evidence from reconstructed historical sequence.
Preserve uncertain paint-transfer interpretations.
Keep current order and original planned order separate.
Use missing leaves only as codicological evidence, not semantic placeholders.
Require decipherments to remain compatible with physical production.
Return every abstract claim to the manuscript image and material object.

What Would Count as a Real Production Breakthrough?

A major advance could come without deciphering a word.

Imagine a complete high-resolution crossing map of every place where outline, text, paint, foliation, marginalia and transfer marks overlap.

The map is coded independently by multiple manuscript specialists.

A stable partial order emerges.

Most bifolia fall into a small number of reproducible production states.

Paint-transfer evidence independently identifies which surfaces faced each other during one colouring stage.

Binding evidence constrains which rearrangements were physically possible.

Now a proposed original ordering can be tested against text statistics and visual continuity without assuming either one first.

That would transform production history from a plausible narrative into a constrained historical model.

The breakthrough would not need to tell us what Voynichese says.

It would tell us which physical states existed when it was being said.

Primary School: Which Mark Came First?

Draw a red line across a blue line.

Make the crossing obvious.

Ask the child which colour is on top.

Then ask:

Do we know how many minutes passed between the two lines?

No.

We know relative order without absolute timing.

That is manuscript stratigraphy in miniature.

Lower Secondary: Build a Partial Order

Give students four operations:

drawing;
writing;
painting;
numbering.

Then give only three clues:

writing avoids drawing;
paint lies over writing;
paint lies over one number.

Ask which orders are forced and which remain uncertain.

The student discovers that evidence can constrain a timeline without uniquely determining every stage.

Upper Secondary: Direct Evidence Versus Reconstruction

Give students two statements.

Paint crosses the folio number at this locus.

The entire manuscript was painted only after all folio numbers were written.

Ask which is direct observation and which is generalisation.

Then ask what additional observations would be needed to justify the generalisation.

This is exactly the reasoning discipline production history requires.

JC and Adult Readers: Production as a Directed Acyclic Graph

At a higher level, manuscript production can be represented as a partial-order graph.

Nodes are operations.

Edges mean “must precede”.

Outline → text.

Text → some paint.

Some numbering → some paint.

Numbering → loss of specified missing leaves.

The graph need not impose an order where evidence does not.

That is its strength.

A narrative wants one timeline.

A partial-order model can preserve uncertainty while still ruling out impossible sequences.

For an object as complicated as Voynich, that may be the more faithful representation.

A Parent and Teacher Guide

Ask which mark lies above which.
Separate relative order from absolute date.
Do not assume one role means one person.
Do not scale one local overlap into a universal timeline automatically.
Keep colour as a later and less secure semantic layer.
Remember that modern pages came from physical bifolia.
Keep current order separate from planned order.
Use reconstructed timelines as hypotheses with explicit confidence levels.

The broader lesson travels far beyond Voynich:

when several historical processes are collapsed into one surviving object, separate the layers before interpreting the whole.

Reader Checklist: Before You Tell the Story of How Voynich Was Made

Is the claim based on a physical crossing or on interpretation?
Is the observation local or manuscript-wide?
Does outline clearly precede text at the locus?
Does paint clearly lie over ink?
Is the paint chronology being generalised beyond the evidence?
Is bifolium structure being preserved?
Are quire and folio numbering treated as separate layers?
Is present order being mistaken for planned order?
Are missing leaves being used only for codicological inference?
Is the present binding being mistaken for the first binding?
Is a tentative reconstruction labelled tentative?
Does the interpretation survive if painting occurred in more than one stage?
Does the decipherment remain valid under page-order uncertainty?

Frequently Asked Questions

Were the Voynich drawings made before the text?

On illustrated pages, the evidence strongly indicates that the principal illustration outlines generally preceded the text, because writing repeatedly avoids and fits around those outlines.

Was the text meaningless filler added around pictures?

Not established. Tight spatial fitting is compatible with generated filler, but also with a prepared draft, abbreviation, flexible notation or another planned image-text workflow.

Was the manuscript painted after it was written?

Paint lies over outlines and apparently over text in many places, so colour is generally a later layer than principal ink work. The full painting chronology may include multiple stages and remains uncertain.

Was one person responsible for drawing, writing and painting?

Unknown. Production sequence establishes roles and operation order more securely than historical identities. One person could perform several roles, or several people could share them.

Was the manuscript made page by page?

The stronger production unit appears to have been the bifolium or foldout sheet. It remains unclear whether makers completed all surfaces in strict sheet-wide stages or worked page by page within each sheet.

Is the current page order original?

Probably not. Several codicological, textual and visual observations strongly suggest that the current order differs from the originally planned sequence.

Has the original order been reconstructed?

No complete reconstruction is accepted. Some local relationships are strong, but the entire original sequence remains unresolved.

When were the folio numbers added?

They are later than the principal page content and were added by a different hand from the quire numbers. Their precise date is less certain than the likely fifteenth-century quire marks.

Why does production order matter for decipherment?

Because a solution should be compatible with the object’s physical history. Theories that require colour to be primary, current order to be original, or missing pages to contain unobserved rescue information carry additional burdens.

What is the strongest current conclusion?

The manuscript preserves a layered production history: outlines, text, paint, organisation and binding are separable operations. Their broad relative order can be constrained much better than the identities, exact timing and complete sequence of the people who performed them.

Research and Further Reading

The Final Idea

We cannot yet read the Voynich Manuscript.

But the manuscript can still tell us which of its own marks arrived first.

The drawing waits beneath the text.

The text bends around the drawing.

Colour crosses both.

Numbering arrives as another layer.

Leaves disappear.

The book is assembled into a state that may no longer preserve its first planned order.

This is not a solution to the mystery.

It is something more fundamental.

It is the object remembering its own construction.

Before we ask what Voynich means, we should listen carefully to what its layers can still tell us about how Voynich came into being.

Voynich | Everything eduKate Knows and Tested | The Order of Making: Drawing, Writing, Paint and Assembly