Voynich Transposition Problem: What If the Symbols Are in the Wrong Order?

Perhaps Voynich contains the right information in the wrong order.

That possibility is older than computers and more dangerous than it sounds.

Substitution hides a message by changing identities.

A becomes X.

B becomes Q.

The sequence remains.

Transposition hides a message by changing sequence.

The same underlying units are rearranged according to a route or permutation.

Write a sentence into a grid.

Read the columns instead of the rows.

Reverse every second group.

Read a circle from a defined starting point.

The plaintext order disappears while its component inventory survives more directly than under substitution.

Voynich practically begs us to try this.

Its exact token-to-token sequence is weaker than many readers expect from ordinary prose.

Its line beginnings and endings behave differently.

Its token edges remain coupled.

Its circular text requires start and direction choices.

Its foldouts create geometry larger than one normal page.

Its present folio order is not guaranteed to be original.

Perhaps we are simply reading in the wrong order.

Then comes the trap.

A six-letter group has 720 possible permutations.

A ten-unit sequence has more than three million.

Add reversal.

Rotation.

Alternative spaces.

Several route shapes.

A desired word can nearly always be found eventually.

A real transposition theory must specify how the manuscript is rearranged before the resulting plaintext is allowed to tell us whether the route was convenient.

That is the Transposition Problem.

Quick Read

Direct answer: transposition and route-based reading are legitimate mechanism classes for testing Voynich, especially because sequence structure varies by resolution and because circular, radial and foldout text has genuine geometry. But no accepted fixed transposition route has recovered a source language from the manuscript. Every proposed rearrangement must be physically or cryptographically motivated in advance, preserve a compact rule across unseen passages, restore stable source grammar, and explain rather than merely exploit line, edge, Currier and document-role structure.

Substitution changes symbol identity; transposition changes order.
A system can combine both.
Transposition can occur within words, among groups, among words, across lines or through a geometric route.
Reversal is the simplest possible transposition.
Columnar or route reading changes adjacency while preserving the underlying inventory more strongly.
Voynich ordinary prose is visibly written left-to-right, which constrains but does not prove semantic reading order.
Circular text adds rotation and direction freedom that must be controlled explicitly.
Radial text can run inward or outward and should not be rearranged semantically without local evidence.
Current page order and ciphertext transposition are different problems.
Physical rebinding can reorder pages without implying that text inside each page was enciphered by transposition.
Weak exact whole-token dependence can be compatible with transposition, sparse vocabulary or the wrong unit of analysis.
Strong edge dependence survives as a major constraint any transposition must reproduce.
Free anagramming is not decipherment.
Every optional reversal, rotation and permutation is model freedom and must be counted.
A valid route should recover source morphology and syntax on unseen text without being redesigned.
No accepted Voynich transposition key or route currently meets that burden.

Substitution and Transposition Ask Different Questions

Suppose the plaintext is:

MEDICINA

A substitution cipher may produce:

XQPLZPTN

The order positions remain linked.

A transposition might produce:

DMIIAENC

The symbols are the same plaintext letters but reordered.

A combined cipher can first substitute, then transpose.

Voynich therefore cannot be tested against one vague category called “cipher”.

Different transformations destroy different source statistics.

Simple Reversal Is the Cheapest Test

Before inventing complex routes, reverse the sequence.

Whole text.

Individual tokens.

Lines.

A reversal swaps prefixes and suffixes.

Token-initial q becomes token-final.

Minim endings become beginnings.

Boundary directionality reverses.

Voynich positional grammar is strongly asymmetric.

That makes arbitrary reversal expensive.

A reversed plaintext model must explain why the ciphertext’s visible forward direction creates such coherent internal positions.

Anagramming Each Token Is Almost Unlimited Freedom

One common decipherment temptation is local rearrangement.

Take one Voynich token.

Assign tentative letters.

Reorder them until a familiar word appears.

This method can feel productive because every token becomes a small puzzle.

Statistically it is dangerous.

Longer words have huge permutation spaces.

Historical spelling adds more variants.

Abbreviation adds more.

Choose among several languages and one attractive match is almost guaranteed.

A serious intraword transposition cipher needs a fixed permutation rule based on length, position, key or another reproducible feature.

A Fixed Intraword Permutation Makes Predictions

Suppose every five-character plaintext word is written in position order:

3-1-5-2-4

Now every five-character token follows one reversible operation.

This is testable.

Apply it to unseen words.

Does the source language become more grammatical?

Do recurring affixes return to stable positions?

If each word length needs a unique hand-picked route, complexity grows.

If each token needs a custom route, the theory collapses into anagrams.

Group Transposition Is More Plausible Than Arbitrary Letter Shuffling

Voynich glyphs may form multi-symbol units.

Bench groups.

Minim endings.

q-o openings.

If those are functional chunks, a cipher could reorder chunks rather than individual glyphs.

This preserves internal unit structure while changing larger order.

The idea has one advantage.

It respects Voynich’s strong internal token grammar better than free character scrambling.

Its burden is identifying the chunks independently of the desired plaintext.

Word-Level Transposition Could Hide Syntax

Move whole words rather than letters and word forms remain intact while syntax disappears.

This is tempting because Voynich exact word-order signals can look weak under some measurements.

But weak exact-token dependence has alternative explanations.

large sparse vocabulary;
wrong word boundaries;
grammar living at word-family or edge level;
record-like rather than sentence-like text;
cipher transposition.

Word-level transposition therefore remains one model among several.

Its distinctive prediction is that one fixed inverse ordering should restore substantially stronger source syntax.

Reddy & Knight Make the Order Question Measurable

Computational work on Voynich has long examined sequence dependencies without needing semantic values.

Reddy and Knight found relatively weak exact whole-word order in one Currier-B sample compared with controls, but they also detected character-edge relationships across word boundaries and latent word classes.

This combination matters.

A transposition that completely randomises word order should destroy many boundary relationships.

Voynich retains some.

Any rearrangement model therefore needs to preserve or recreate the dependencies that survive.

The 2026 Edge Upgrade Raises the Bar Further

Recent preprint work reports an especially sharp asymmetry:

exact token identity predicts the next token weakly, while glyphs touching token boundaries remain more strongly coupled.

A word-level transposition might explain weak whole-token order.

It must then explain why the boundary edges still know about one another.

Perhaps transposition operates above the edge grammar.

Perhaps visible tokens are not source words.

Perhaps no transposition is required.

The new discriminator is useful precisely because several mechanisms that fit old word-order intuitions now make different edge predictions.

Line-Level Routes Are Tempting Because Lines Are Strong Units

Voynich lines have distinctive beginnings and endings.

This invites route ciphers.

Read odd lines left-to-right and even lines right-to-left.

Read first words down the margin, then return for the rest.

Interleave adjacent lines.

Again, each is testable if chosen from physical evidence.

They are worthless if selected because one route happens to create a desired phrase.

The line must supply the route before the plaintext does.

Could First Characters Have Been Written as a Separate Stream?

Some Voynich passages have prompted the suggestion that first characters of several lines may have been written before line interiors were completed.

If true locally, production order differs from final reading order.

That is important.

It does not automatically make the text a transposition cipher.

A scribe can establish margins first for layout reasons and still intend ordinary left-to-right reading.

Writing order and semantic reading order are separate variables.

Circular Text Creates Rotation Freedom

A circle has no natural first character.

If the writing runs clockwise, the sequence is known only up to rotation unless a start marker exists.

Reverse direction and another full set of sequences appears.

This resembles a transposition search problem even when no cipher exists.

A theory can rotate a ring until a star name appears.

Then reverse it if that works better.

This is precisely why the Direction article insists that physical start and writing direction be fixed before semantic matching.

Radial Text Creates a Different Route Problem

Voynich radial loci can run inward or outward.

The direction is local.

A proposed transposition that reads all radial text toward the centre for semantic reasons would conflict with the visible writing orientation of many loci.

Again, production evidence constrains permissible routes.

Foldouts Invite Route-Cipher Fantasies

A large foldout looks like a map.

Or a table.

Or a route surface.

Perhaps text must be followed across panels in a special order.

Possible.

But folds are physical technologies first.

Creases do not automatically define semantic cells.

Some writing crosses them.

A route theory must derive the traversal from diagram or text structure rather than from the fact that the page is unusual.

Page Reordering Is Not Text Transposition

The Voynich codex has been rebound and rearranged.

Some current page adjacencies are not original.

This absolutely affects manuscript interpretation.

But it is a codicological order problem.

Not automatically an encryption algorithm.

A physical binder can move bifolia after the text is complete.

A transposition cipher intentionally rearranges message units under a key or rule.

Conflating them makes accidental history look like deliberate cryptography.

Original-Order Reconstruction Can Still Help Test Transposition

Suppose a transposition theory works only under current page order.

If codicology shows the relevant sheets were once elsewhere, the theory weakens.

Suppose instead the route follows physical bifolia and survives reconstructed original order.

That is stronger.

Physical manuscript history can therefore act as an independent constraint on route models.

A Transposition Cipher Must Preserve an Executable Procedure

The intended reader must know what to do.

Reverse every word?

Read columns under a keyword?

Take alternate groups?

Follow numbered cells?

A real system must be reproducible by a human operator.

Voynich’s almost-correctionless execution means any route should be simple enough to apply reliably across a long codex.

Free Transposition Is More Dangerous Than Free Substitution

Give one symbol five possible plaintext values and the model becomes flexible.

Give ten symbols arbitrary order and the flexibility explodes factorially.

This is why transposition claims require especially severe control of search freedom.

How many routes were tested?

How many reversals?

How many starting points?

How many segmentation choices?

One impressive plaintext found after millions of candidate permutations has little evidentiary force unless the search multiplicity is accounted for.

The Route Must Be Chosen Without Reading the Target Language

This is the golden rule.

Use physical geometry.

Repeated markers.

A documented historical cipher rule.

A stable numeric key.

Anything independent.

Do not choose the route because Latin appears after trying it.

That is post-selection.

The route needs an existence before the plaintext confirms it.

A Real Transposition Should Restore Source-Side Regularity

Undo the correct transposition and source order should return.

function words recur naturally;
morphology occupies stable positions;
phrases repeat at plausible rates;
syntax strengthens;
proper names retain consistent spelling;
paragraphs become coherent source units.

If a route produces one readable word but makes surrounding structure worse, it is not enough.

The correct ordering should improve the whole system.

Transposition Alone Cannot Explain Everything Voynich-Like

Pure transposition preserves the source symbol inventory and frequency more strongly than substitution.

Voynich’s character system is itself unusual.

Its token-internal grammar is severe.

Its visible alphabet does not look like straightforward rearranged Latin letters.

Therefore a pure transposition of ordinary Latin or Italian characters is not an economical whole-manuscript model.

A transposition theory will likely require another representation or substitution layer.

Each added layer must earn its complexity.

Combined Substitution and Transposition Can Hide More—but Costs More

First substitute.

Then rearrange.

Now both identity and order are hidden.

This can explain more surface anomalies.

It also creates more free parameters.

A serious hybrid theory should outperform simpler homophonic, codebook and generator controls enough to justify the extra route layer.

Complexity is acceptable when it buys unique predictive power.

Complexity added after each failure is not.

The Crib Problem Becomes Even Harder

A proper-name crib normally aligns one expected string to one visible string.

Transposition breaks that alignment.

The letters of the expected name may appear in another order.

Now a solver can permute the ciphertext until the desired name emerges.

This dramatically increases false-match risk.

A transposed crib must therefore have its route fixed independently before the name is tested.

The Numbers Problem Could Supply a Route—If It Ever Becomes Secure

Numbers are natural route keys.

Read cells 3-1-4-2.

Follow thirty zodiac positions from a known origin.

Use a sequence around a wheel.

Voynich contains countable geometry but no accepted principal-script numeral system.

Therefore numerical route hypotheses remain attractive but unanchored.

If a secure number system is ever recovered, transposition should be retested immediately.

f57v Could Become a Route Key Only After Its Function Is Known

Repeated ordered signs on f57v make it look like something a route cipher could use.

Alphabet order.

Permutation sequence.

State table.

But that is the same temptation seen in the Codebook Problem.

The ring is structurally real.

Its cryptographic function is not.

Using it as the route key before independently identifying what the sequence represents risks fitting the key to the desired outcome.

Currier States Could Be Route States

Perhaps A and B use different permutation rules.

This is possible.

It is also expensive.

Every additional route state adds key information that historical users must know.

A strong theory should show that the A→B change point predicted by the route aligns with independent Currier or production evidence.

If route states are invented after seeing text clusters, they merely rename the clusters.

The Drift Problem Could Falsify Abrupt Route Changes

If Voynich changes gradually rather than through sharp key switches, a transposition theory built on abrupt A/B permutations may struggle.

Conversely, a continuously changing route would be operationally difficult for a historical reader.

Textual drift therefore constrains plausible key-state architecture.

Again, one branch of evidence can defeat another branch’s convenient freedom.

The Almost-Correctionless Manuscript Is a Human-Factors Test

Complex transposition is cognitively expensive.

A writer must track position.

Store source material.

Apply the permutation.

Maintain line and paragraph geometry.

Voynich shows few visible corrections.

This favours simple, practised or externally supported routes over elaborate mental shuffling.

A cipher must fit the hand that actually had to write it.

A Good Transposition Model Should Be Reversible Forward

Once a route is proposed, take period-appropriate plaintext.

Apply the route forward.

If substitution or homophony is also claimed, apply those exact fixed rules too.

Does the resulting ciphertext reproduce:

word-length structure;
edge coupling;
line effects;
exact repetition;
Currier regimes;
label/prose differences?

Forward construction is one of the strongest guards against post-hoc rearrangement.

What Survives the Transposition Work

Transposition is a legitimate cipher mechanism distinct from substitution.
Voynich has genuine order ambiguities in circular, radial and physically rearranged contexts.
Weak exact token-order signals make sequence transformation worth testing.
Strong positional and edge signals severely constrain arbitrary reordering.
Intraword, group, word, line and route transposition make different predictions.
Current page order uncertainty is codicological and should not be confused with intentional ciphertext transposition.
Every optional permutation, rotation and reversal adds large model freedom.
A real transposition route must be fixed before semantic fitting.
The inverse route should restore stable source-language structure across unseen text.
No accepted Voynich transposition route currently does so.

What Does Not Survive as Established Knowledge

Voynich words are proven anagrams.
Every line should be read backwards.
Circular text may be rotated until a desired name appears.
Foldouts are proven route-cipher grids.
Current misbinding proves intentional textual transposition.
f57v is a proven permutation key.
Weak exact token syntax proves word-order encryption.
A readable word found after free rearrangement validates a route.
Transposition alone explains the Voynich character system.

Order remains a legitimate unknown.

The Voynich permutation does not.

A Better Transposition Analysis

Specify the level being permuted: character, group, token, line or spatial cell.
Derive the route from physical, cryptographic or independently recovered key evidence.
Count every tested reversal, rotation and permutation.
Freeze the route before source-language evaluation.
Apply it to unseen text.
Require stronger source phonotactics, morphology and syntax.
Preserve edge and boundary evidence rather than scrambling it away.
Separate page-order reconstruction from ciphertext transposition.
Test human executability against the clean correction profile.
Encrypt forward under the same route.
Compare with simpler models and penalise extra route complexity.
Keep every failed fixed route visible.

What Would Count as a Real Transposition Breakthrough?

Imagine one recurring geometric or textual marker independently identifies a four-step reading route.

The route is frozen without reference to any candidate plaintext.

Applied to one development corpus, it increases source-like sequence structure.

Then it is applied unchanged to unseen bifolia.

Function-word classes emerge.

Recurring morphology returns to stable positions.

Proper-name cribs align without choosing new rotations.

Currier differences become explainable under one compact key-state rule.

An independent operator can take the recovered plaintext and reproduce the observed ordering using the same procedure.

That would be a transposition breakthrough.

The correct Voynich order, if it is hidden, will earn itself by restoring many independent structures under one route—not by letting us rearrange each passage until something readable appears.

Primary School: Scramble the Same Letters

Write CAT.

Rearrange it to ACT.

The same letters can form a different word.

Now let the child rearrange three letters freely.

Several results appear.

The exercise shows why finding one meaningful arrangement is not enough; the rearrangement rule must be known.

Secondary School: One Fixed Route

Write a sentence into a four-column grid row by row.

Read it out by columns.

Give the ciphertext to another student along with the rule.

They can recover the message exactly.

Then withhold the rule and allow arbitrary routes.

The difficulty—and risk of false solutions—becomes obvious.

JC and Adult Readers: Search Freedom Is the Real Enemy

At a higher level, transposition analysis is a multiple-hypothesis problem.

Each candidate permutation is a hypothesis.

Search enough permutations and one will maximise any chosen language score by chance.

Therefore significance must account for the full route search, and the final route must be evaluated on untouched text.

Predefined route families and physical constraints are more important than clever optimisation.

A Parent and Teacher Guide

Separate substitution from transposition.
State what level is being rearranged.
Never permit free anagramming as evidence.
Use physical writing direction and geometry to constrain routes.
Keep page reordering separate from cipher transposition.
Count every tested rotation and reversal.
Freeze the route before translation.
Demand source grammar and unseen-text prediction.

The transferable lesson is:

when rearrangement is allowed, a readable result is cheap; an independently justified route that keeps working is expensive evidence.

Reader Checklist: Before You Accept a Voynich Transposition

What units are permuted?
What defines the route?
Was the route chosen before seeing plaintext?
How many routes were searched?
Are reversals allowed?
Are circular rotations allowed?
Are spaces movable?
Does the route respect visible writing direction?
Does it respect token-edge structure?
Is current page order being confused with encryption order?
Does the inverse route improve source syntax?
Does it work on unseen bifolia?
Does it require different routes by Currier or hand?
Can a historical operator execute it reliably?
Can the recovered plaintext be encrypted forward under the same route?

Frequently Asked Questions

What is a transposition cipher?

It conceals plaintext by rearranging the order of units under a rule or key rather than only replacing each unit with another symbol.

Could Voynich words be anagrams?

They can be tested under fixed intraword permutations, but free anagramming has too many possibilities to provide strong evidence.

Does weak word order prove transposition?

No. Sparse vocabulary, wrong token units, record-like documents and class-level grammar can also weaken exact token-to-token prediction.

Does the manuscript’s misbinding prove text transposition?

No. Rebinding is physical history. A transposition cipher is an intentional message transformation. The two must be analysed separately.

Could circular text hide a transposition?

Possibly, but start point and direction must be fixed from physical or structural evidence before semantic matching, otherwise rotation and reversal create many false alignments.

What is the strongest current conclusion?

Order remains a legitimate unresolved dimension of Voynich, but its strong internal positional and boundary structure constrains transposition sharply. No accepted fixed route has recovered a stable source text.

Research and Further Reading

The Final Idea

Voynich has several real order problems.

Which way around a circle?

Which way along a radius?

Which bifolium came first?

Where does one sentence end?

Those uncertainties make transposition tempting.

They also make it dangerous.

Every unknown order is another lever a solver can pull until something looks readable.

The discipline is to lock the levers before reading the answer.

If Voynich is transposed, the right order will not be the order that gives us one beautiful sentence. It will be the one physically or cryptographically justified route that makes thousands of independent structures fall into place without asking us to rearrange them again.

Voynich | Everything eduKate Knows and Tested | The Transposition Problem: What If the Right Symbols Are in the Wrong Order?