Voynich Direction Problem: Circular and Radial Writing

Most writing gives us a quiet instruction before we read a single word.

Start here.

Move this way.

Continue until the line ends.

Then move to the next line.

In the ordinary Voynich paragraphs, that instruction is surprisingly familiar. The writing proceeds overwhelmingly from top to bottom and from left to right.

Then the manuscript enters a circle.

And suddenly “next” becomes a much harder word.

Writing travels around circumferences. Other strings point inward along radii. Others point outward. Start points are sometimes visually marked. The Rosettes foldout contains circular text that changes orientation around a full turn, and one small circular locus is documented as running anticlockwise.

Direction, in other words, is not decoration.

The Voynich Manuscript does not merely contain text inside diagrams. In some places, the geometry of the diagram becomes part of how the text must be traversed.

That creates a research problem before translation begins.

Quick Read

Most ordinary Voynich paragraph text is written top-to-bottom and left-to-right.
Circular, astronomical and cosmological pages contain text written around circumferences and along radii.
Radial writing can point inward or outward depending on the locus.
Several diagrams contain visually indicated start points for circular writing.
On the Rosettes foldout, circular writing rotates around the page so that its physical orientation changes continuously with the circle.
One small Rosettes circular text is documented as anticlockwise, showing that “all circular Voynich text is clockwise” is too strong.
A modern transcription must linearise spatial writing, which means it must choose a start point, direction and order.
Those choices can change adjacency statistics even when every glyph has been copied correctly.
Direction can therefore be evidence about document use, production and information architecture without being a translation.
A strong decipherment should work with the manuscript’s spatial order rather than relying on one arbitrary linearisation.

The Ordinary Case Is Useful Because It Gives Us a Baseline

René Zandbergen’s descriptive survey makes the ordinary writing direction clear: most Voynich text is written line by line, top to bottom, left to right.

That baseline matters.

If the whole manuscript were written in a radically unfamiliar directional system, every orientation decision would be uncertain.

Instead, ordinary paragraphs give us something stable. A line has a left edge and a right edge. Line beginnings and endings can be studied statistically. Paragraph starts can be compared. Left margins can be observed.

The dedicated Line as a Unit article owns those ordinary line-position effects.

This article begins where that baseline stops being sufficient.

A Circle Has No Natural First Word

Take a sentence printed in one horizontal line.

The line tells you where to begin because the writing convention and page geometry agree.

Now bend the same sentence around a circle.

The circle itself has no privileged beginning.

You need another cue.

a gap;
a vertical marker;
a nearby label;
a change in orientation;
a conventional top position;
a relation to the central image;
or prior knowledge of the diagram’s structure.

Several Voynich circular texts appear to provide start cues. On f57v, for example, four concentric bands have marks aligned near a common starting region. Other circular pages have large gaps that suggest where a text item begins.

That is important because the manuscript is not merely wrapping prose for decoration.

It is managing an orientation problem.

Clockwise Is Not the Same as Left-to-Right

This sounds trivial until a transcription system has to encode it.

Imagine writing clockwise around a circle while keeping every word locally readable from left to right along the tangent.

At the top of the circle, the line may feel roughly horizontal.

Along the right side, the writing becomes vertical.

At the bottom, it is upside down relative to a stationary reader.

Along the left side, it becomes vertical again in the opposite orientation.

This means two coordinate systems coexist:

local writing direction — how characters follow one another within the string;
global path direction — how that string travels around the diagram.

A digital text file usually preserves the first and destroys the second unless the metadata explicitly records it.

The Rosettes Foldout Makes Direction Physical

The Rosettes foldout is the extreme case because it refuses to behave like an ordinary page.

Its nine major circular regions are connected across a large folded sheet. Numerous labels and floating passages sit around them. Circular text follows several perimeters.

Zandbergen’s folio-level description notes that the writing around the major circles turns through a complete revolution. On the left side of a circle, a writer can move in one physical orientation; near the top the page relation changes; on the right the hand must approach the surface differently; near the bottom it changes again.

He cautiously suggests that this physical arrangement may be consistent with a right-handed writer working with the sheet positioned to the writer’s right.

That is a production hypothesis, not a semantic interpretation.

And it is a useful example of what direction can reveal.

Orientation can tell us something about how a page was handled even when it tells us nothing about what the sentence means.

See The Rosettes Foldout and The Foldouts.

The Anticlockwise Exception Matters More Than Its Size

One small circular text on the Rosettes sheet is described as the manuscript’s only documented anticlockwise circular writing.

The text is short.

The methodological consequence is large.

It means a rule such as “Voynich circular text always runs clockwise” is already falsified by the object.

That forces better questions.

Is the anticlockwise locus a different document role?
Was the sheet rotated differently during writing?
Does the direction follow a local visual relationship?
Is the reading direction itself part of the diagram?
Could the apparent exception reflect how the circular locus was segmented?

The correct response to an exception is not to erase it.

It is to make the rule smaller and more accurate.

Radial Writing Creates a Different Problem

A radius has two natural directions.

Centre → edge.

Edge → centre.

The Voynich Manuscript uses both.

On f67r1, twelve radial text items are documented as running inward. On f67v1, seventeen radiating text lines run outward. On f67v2, the eight radial loci split: four inward, four outward. On f69r, twenty-two radial text items run outward; on f69v, twenty-eight run inward.

This is not random orientation noise.

The directions occur as organised page-level choices.

That creates a remarkable research question:

Does inward versus outward writing encode relation to the centre, or is it simply the most convenient way to fit text into a particular geometry?

We do not yet know.

But the difference is measurable.

Direction Can Encode Relationship Without Encoding Meaning

Consider a modern diagram.

An arrow pointing toward a centre can indicate input.

An arrow pointing outward can indicate output.

A circular arrow can indicate cycle.

A clockwise order can indicate sequence.

Voynich does not need to use these exact modern semantics for direction to matter structurally.

Direction may indicate:

reading order;
relation to a centre;
temporal progression;
hierarchy;
flow;
diagram construction convenience;
or no semantic value beyond legibility.

The point is not to choose one by intuition.

The point is that any proposed meaning for a radial or circular text should explain why its orientation is what it is.

Direction and the Problem of Linear Transcription

A computer likes a sequence:

token 1 → token 2 → token 3.

A circular diagram may instead contain:

four concentric texts;
several radial texts;
labels;
a central emblem;
start markers;
spatial sectors.

To create a text file, an editor must choose an order.

Which ring first?

Where on the ring does the sequence start?

Clockwise or anticlockwise?

Radial items before labels or after?

Does an inward radial line precede or follow the centre?

Modern IVTFF-style transcription can preserve locus types such as paragraph, label, circular writing and radial writing, which is an important advance because it prevents every string from being flattened into ordinary prose.

Still, every statistical analysis needs to know which order the researcher imposed.

Linearisation is a model of the page, not the page itself.

Why Adjacency Statistics Become Fragile on a Circle

Suppose the last word of a circular transcription sits immediately before the first word physically on the ring.

In the manuscript, they are neighbours.

In a linear file, the first and last items can appear maximally far apart.

Conversely, if a researcher chooses an arbitrary break point, two words placed adjacent in the file may merely be adjacent because of the chosen transcription start.

This matters for:

bigrams;
next-word preferences;
sequence models;
topic locality;
repetition;
candidate grammatical transitions.

For circular loci, the correct mathematical model may sometimes be cyclic rather than linear.

That is not proprietary machinery.

It is simply respecting the shape of the evidence.

Start Markers Are More Valuable Than They Look

A small vertical stroke beside a circular band can seem trivial.

But if it consistently marks where text begins, it can solve one of the hardest problems in circular transcription.

The f57v diagram is particularly valuable because several concentric bands appear to share aligned start cues.

That suggests the rings were not merely filled opportunistically.

They may have been designed as comparable sequences.

The repeated 4×17 character sequence in one ring makes this especially interesting, but the sequence belongs to the character-system problem and the dedicated Key-Like Sequences article, not to an automatic “alphabet wheel” interpretation.

Does Direction Tell Us Which Hand Wrote the Page?

Sometimes physical orientation can suggest writing posture.

A right-handed writer may prefer particular rotations when following a circle.

But this inference is delicate.

The parchment can be rotated.

A foldout can be approached from different sides.

A writer may stand or sit.

The text may have been added in several sessions.

Direction should therefore support palaeographic hypotheses, not replace palaeography.

See The Scribe Problem.

Could Direction Encode Time?

Circular diagrams naturally invite temporal interpretation because calendars and astronomical cycles repeat.

Clockwise movement can feel like progression.

Inward movement can feel like convergence.

Outward movement can feel like unfolding.

Those metaphors are cognitively strong.

They are not evidence.

A temporal interpretation would need independent support from counts, repeated labels, known celestial anchors or cross-page correspondences.

Direction can constrain the model.

It cannot name the time unit by itself.

Could Direction Encode Flow?

The same warning applies to flow.

Voynich contains tubes, pools, connectors and radial structures.

It is tempting to make inward and outward writing into input and output.

That could be correct in some local diagram.

But a reading-direction arrow is not the same thing as a depicted fluid-flow arrow.

The manuscript rarely gives us modern arrowheads.

Therefore the analyst must separate:

direction of writing;
direction of drawing;
direction of depicted process;
direction of reading.

They may coincide.

They do not have to.

What the Direction Problem Gives Us

A strong baseline: ordinary paragraphs are top-to-bottom and left-to-right.
A second mode: circular writing integrated into diagrams.
A third mode: radial writing that can run inward or outward.
Evidence that some circular loci have explicit or implicit start points.
An anticlockwise exception that prevents an over-simple universal clockwise rule.
A reminder that spatial text cannot be reduced to a flat corpus without metadata loss.
A production clue: page rotation and hand posture may sometimes be inferable cautiously.
A methodological rule: sequence claims must respect the original geometry.

Primary School: Put a Sentence on a Circle

Write eight words on separate cards.

Put them in a straight line.

The first word is obvious.

Now arrange the same cards in a circle.

Ask:

Where do we start?
Which way do we go?
What happens after the last card?

The child immediately discovers why geometry can be part of reading.

Secondary School: Linearise the Same Diagram Two Ways

Take a circular sequence of twelve labels.

Create two transcriptions.

Start at the top and read clockwise.
Start at the bottom and read clockwise.

The words are identical.

The linear bigrams at the break point are different.

This demonstrates how a representation decision can alter statistics without altering the manuscript.

JC and Adult Readers: Treat Spatial Text as a Graph Before a String

A circle is naturally cyclic.

A radial diagram is naturally relational.

Before imposing a single sequence, record:

locus identity;
start marker;
clockwise or anticlockwise path;
inward or outward radial orientation;
centre association;
ring membership;
physical neighbours.

Only then choose a linear form for the particular analysis.

This protects the page from the convenience of the spreadsheet.

Reader Checklist: Before You Analyse Circular Voynich Text

Is the locus ordinary prose, label, circular text or radial text?
Is a start point visible?
Is direction clockwise or anticlockwise?
If radial, does it run inward or outward?
Does the transcription preserve this metadata?
Is the final word physically adjacent to the first?
Was the page probably rotated during writing?
Could orientation be driven by legibility rather than semantics?
Does another comparable diagram use the same directional rule?
Would the result change under a different legal linearisation?
Is a direction claim being confused with a flow claim?
Does the proposed interpretation predict orientation elsewhere?

Frequently Asked Questions

Is Voynichese written left-to-right?

Ordinary paragraph text is overwhelmingly written top-to-bottom and left-to-right. Diagrammatic text introduces circular and radial orientations that need separate treatment.

Is all circular Voynich text clockwise?

No. Most documented circular loci follow the dominant direction, but one small circular text on the Rosettes foldout is specifically described as anticlockwise.

Does radial text run outward?

Sometimes. Other radial loci run inward, and f67v2 contains both inward and outward radial text on the same diagram.

Why does direction matter for decipherment?

Because sequence-based analyses depend on which item follows which. A circular or radial page can be misrepresented if its spatial ordering is flattened incorrectly.

Can direction reveal meaning?

Direction can constrain relationships and reading order, but no general semantic meaning for clockwise, anticlockwise, inward or outward writing has been established.

Research Foundations

The Final Idea

Most books tell the reader where to go next with lines.

Voynich sometimes tells the reader with geometry.

That may sound like a small observation.

It changes the unit of analysis.

A ring is not merely a bent paragraph.

A radius is not merely a vertical line.

Direction can be part of the information architecture even while the words remain unread.

Before asking what a Voynich sequence says, first make sure we know which way the sequence goes.

Voynich | Everything eduKate Knows and Tested | The Direction Problem: Left-to-Right, Clockwise, Anticlockwise and Radial Writing