Voynich Shorthand Problem: Could It Be Tachygraphy?

Perhaps Voynich is not hiding a message behind a cipher.

Perhaps it is writing the message directly—but in a compression system we no longer know how to read.

That is the shorthand hypothesis.

And it is more interesting than the word “shorthand” first suggests.

Modern shorthand evokes fast note-taking.

Curves.

Hooks.

Compressed syllables.

Medieval and early modern shorthand traditions could be much stranger.

One sign might encode a syllable.

Another a common word.

A final stroke might alter a vowel or ending.

Forms could be learned through convention rather than spelled alphabetically.

That kind of writing can look opaque to anyone outside the tradition while remaining ordinary, meaningful notation to a trained user.

The Voynich Manuscript has attracted shorthand explanations before.

Famously, William Romaine Newbold claimed in the early twentieth century that microscopic details inside Voynich characters concealed a shorthand system encoding Roger Bacon’s scientific discoveries.

That solution collapsed.

The “micrographic” marks were largely ordinary irregularities, and Newbold’s decoding procedure gave itself enough interpretive freedom to manufacture readings.

That historical failure matters.

But it does not mean every future shorthand hypothesis is automatically Newbold.

In March 2026, with a revised version in April, Matt Ruckman released a preprint proposing Italian syllabic tachygraphy as the mechanism most consistent with several statistical properties of Voynichese.

The revised paper reports a simultaneous match on three diagnostics under its chosen representation: entropy shift, cross-boundary mutual information and frequency-connectivity structure.

It also reports partial lexical signals and several falsifiable predictions.

Crucially, the paper itself states that this is hypothesis development, not a decipherment.

That is exactly how the claim should be treated.

The Shorthand Problem is not whether one statistical model can make Voynich resemble one historical notation. It is whether a historically documented shorthand system can explain the manuscript’s character inventory, token structure, boundaries, Currier variation and unseen text under one fixed expansion grammar.

Quick Read

Direct answer: shorthand/tachygraphy is a legitimate Voynich mechanism class distinct from simple alphabetic abbreviation and from classical ciphertext. A 2026 preprint by Matt Ruckman proposes Italian syllabic tachygraphy and reports several encouraging statistical matches, but it remains unreviewed hypothesis development, covers only part of the proposed sound/value inventory, has not produced connected accepted plaintext, and must survive independent replication before it can change the manuscript’s status.

Tachygraphy is a system for compressed writing, often using conventional signs for syllables, words or recurring sound groups.
It can be meaningful writing without being encryption.
It can also be deliberately obscure to outsiders because the conventions require training.
Shorthand differs from ordinary abbreviation because the whole writing system may operate at a different unit scale.
Newbold’s famous shorthand claim failed because its microscopic premises and interpretive procedure were not robust.
That failure does not logically rule out all historically grounded shorthand mechanisms.
Ruckman’s 2026 preprint proposes Italian syllabic tachygraphy associated with northern Italian notarial practice.
The April revision reports a simultaneous match on three statistical diagnostics rather than relying on one entropy pattern.
The preprint compares its mechanism with several alternatives, including Naibbe-like verbose substitution, self-citation and Cardan-grille-style generation.
It reports partial syllable values and word-level signals, but also a major coverage limitation.
The author explicitly describes the work as hypothesis development rather than decipherment.
The proposed shorthand must still explain Currier A/B, labels, line effects, word-family structure, rare glyphs and the physical scribal system.
Any partial lexical readings must be protected against dictionary-collision and multiple-testing effects.
A mature shorthand model should expand unseen Voynich into coherent source-language morphology and syntax with decreasing ambiguity.
No accepted Voynich shorthand decipherment currently exists.

Shorthand and Abbreviation Are Related but Not Identical

A normal medieval abbreviation starts with ordinary writing and compresses selected pieces.

A bar means omitted letters.

A special sign represents a common ending.

A suspension drops the rest of a word.

Tachygraphy can go further.

The basic unit itself may be syllabic or highly conventional.

A sign’s shape may not transparently resemble alphabetic spelling.

Therefore the existing Abbreviation or Alphabet article owns ordinary scribal compression.

This article asks whether the entire principal script could belong to a shorthand-like notation technology.

A Shorthand System Changes the Unit of Analysis

If one visible sign represents a syllable, character entropy measured as though each sign were a letter can look unusual.

If one visible token contains several shorthand units, word length no longer corresponds directly to source spelling length.

If spaces separate shorthand phrases rather than lexical words, token statistics become distorted.

This is why shorthand is an inverse-representation hypothesis.

It says the manuscript may be orderly at a different linguistic resolution than the one our transcription assumes.

Why Italian Notarial Tachygraphy Enters the Discussion

The 2026 preprint focuses on a historical shorthand tradition documented in northern Italian notarial contexts.

That matters because the Voynich parchment dates to an early-fifteenth-century material horizon and several independent historical hypotheses already place possible production contexts in northern Italy.

But geographic compatibility is not provenance.

A shorthand tradition existing in northern Italy does not prove Voynich was written there.

Nor does statistical similarity prove lineage from that tradition.

The historical control answers only:

could a writing technology of this general kind have existed near the right time and place?

Yes is useful.

It is not identity.

The 2026 Three-Diagnostic Claim

Ruckman’s revised April 2026 paper deliberately strengthens its standard beyond one statistic.

It reports that its tachygraphic representation simultaneously reproduces three Voynich-related signatures:

an entropy-shift pattern;
cross-boundary mutual information;
a frequency-connectivity correlation related to word-family structure.

This is methodologically better than claiming victory because one histogram looks similar.

The paper also reports that several tested alternatives can reproduce one signature while failing others.

That is the right direction for Voynich research.

Joint constraints discriminate better than isolated resemblance.

But the result remains a preprint produced by the proponent of the model.

Independent replication is essential.

The Paper Also Reports Partial Lexical Signals

The revised preprint reports a set of statistically filtered candidate words and a smaller group of content identifications associated with pharmaceutical Latin.

Those claims are intriguing.

They are also exactly where decipherment research becomes most vulnerable.

A large dictionary gives many opportunities for accidental matches.

Flexible segmentation gives more.

Multiple candidate sound values give more again.

Ruckman’s own later writing on the “dictionary collision effect” emphasises this danger.

That self-critique is valuable.

The correct public standard remains:

candidate words become evidence only when one frozen shorthand table predicts connected unseen text more strongly than dictionary search predicts coincidences.

Coverage Is the Honest Limitation

The 2026 preprint explicitly reports incomplete coverage.

Its confirmed proposed syllable values explain only a minority of ordinary Latin text under the paper’s own comparison.

That is not a footnote.

It is the central unresolved engineering problem.

Connected plaintext requires enough of the value system to decode grammar, not just scattered signals.

A model can be statistically suggestive while still far from operational decipherment.

Newbold Is the Historical Warning

Newbold also believed Voynich concealed shorthand.

That superficial similarity must not be ignored.

His method depended on interpreting microscopic irregularities inside normal-looking glyphs as hidden shorthand strokes and then manipulating those readings further.

The solution produced extraordinary Baconian science.

It did not survive scrutiny.

The modern shorthand hypothesis differs if—and only if—it remains tied to visible reproducible glyph units, documented historical comparators and fixed statistical predictions.

The lesson from Newbold is not “never test shorthand”.

It is “never let microscopic or lexical freedom become the decoder”.

Shorthand Must Explain the Alphabet Problem

How many functional shorthand signs exist?

Are benches one sign or several?

Are gallows modifiers?

Are minim strings independent syllables or parts of a larger form?

A shorthand solution should make these segmentation questions easier.

If every disputed glyph can be split differently depending on the target word, the model repeats the old flexibility problem.

Shorthand Must Explain the Vowel Problem

Syllabic shorthand naturally changes how vowel information appears.

A consonant-vowel syllable may be one sign.

Vowel distinctions may be encoded by shape variants or context.

This makes shorthand an attractive explanation for the absence of a clean visible vowel class.

But the proposed expansion must reconstruct a stable vowel system rather than simply assert that shorthand hides it.

Shorthand Must Explain Phonotactics

The newly published Phonotactics Problem becomes the downstream test.

Expand the shorthand.

Do legal Italian or Latin syllable sequences emerge?

Do recurring grammatical endings stabilise?

Do candidate words obey the same sound rules?

If each lexical match needs a different expansion convention, the shorthand table is not converging.

Shorthand Must Explain Word Boundaries

A shorthand tradition may not place gaps exactly where ordinary spelling places them.

That makes the recent Edge Problem relevant.

Some uncertain Voynich spaces may divide larger shorthand units.

Or boundary signs may encode syllable transitions.

A tachygraphic model should predict which gaps are weak and which are strong from its own unit grammar.

This is stronger than adjusting spaces after failed decoding.

Shorthand Must Explain Currier A and B

If Voynich is one shorthand system, why do A and B differ so much?

Possible answers include:

different shorthand dialects;
different scribal conventions;
different source-language registers;
different syllable preferences;
different underlying content.

A serious tachygraphic theory must choose among these with evidence.

It cannot let “shorthand variation” absorb every Currier difference without a rule.

Shorthand Must Explain the Scribes

A trained shorthand system can be written fluently.

That fits Voynich’s relatively smooth production better than a cumbersome improvised code.

But multiple proposed hands create another test.

Do all hands use the same shorthand values?

Do they differ only in ductus and allography?

Or does each hand require a new decoding table?

A shared learned notation should transfer substantially across writers.

Labels Are a Severe Test

If labels are names, shorthand may encode them syllabically.

That could make them look unlike ordinary spelling while remaining pronounceable after expansion.

A mature shorthand model should therefore decode labels with the same table.

If prose values cannot be transferred to labels, either labels use another register or the model is incomplete.

The Almost-Correctionless Manuscript Is a Point in Shorthand’s Favour—But Only a Point

Shorthand is designed for fluent human execution.

A trained writer can produce compressed forms rapidly without constant reference to a key.

This is operationally attractive for Voynich.

Yet fluent handwriting is compatible with language, cipher, notation and generation too.

Production ergonomics keep shorthand viable.

They do not identify it.

The Historical Tradition Must Match More Than Shape

Two writing systems can both use hooks and loops without sharing ancestry.

Therefore visual resemblance to tachygraphic specimens is weak by itself.

Stronger evidence would combine:

stroke construction;
functional sign inventory;
syllable or abbreviation values;
positional behaviour;
historical chronology;
geographic transmission;
predictive decoding.

Functional convergence matters more than a handful of lookalike characters.

The 2026 Model Needs Independent Replication

This cannot be skipped.

Reimplement the proposed representation from the paper.

Use independent Voynich transcription choices.

Recompute the three headline diagnostics.

Audit parameter choices.

Repeat the dictionary-collision controls.

Test held-out folios selected before value tuning.

A preprint becomes durable evidence when other researchers can reproduce both its successes and its failures.

What Survives the Shorthand Work

Historical shorthand and tachygraphy are real writing technologies distinct from simple substitution cipher.
A syllabic shorthand can naturally alter visible entropy, word length, vowel structure and boundaries.
Newbold’s famous shorthand decipherment failed and remains an important warning against flexible hidden-feature interpretation.
Ruckman’s 2026 preprint provides a new, quantitatively explicit Italian syllabic tachygraphy hypothesis.
The revised paper reports simultaneous agreement on several statistical diagnostics rather than one resemblance.
It also reports partial lexical signals and explicit limitations.
The work is not an accepted decipherment and has not produced connected consensus plaintext.
Independent replication remains essential.
A real shorthand model must explain Currier, hand, label and boundary variation under a compact table.
Coverage and dictionary-collision risk remain major constraints.

What Does Not Survive as Established Knowledge

Voynich is proven Italian shorthand.
The 2026 preprint has deciphered the manuscript.
Reported candidate Latin words are accepted plaintext.
Statistical similarity proves direct historical descent from one shorthand tradition.
North-Italian shorthand proves Paduan provenance.
Newbold’s failure proves all shorthand hypotheses impossible.
Every unclear space may be merged to improve a shorthand reading.
Every glyph may receive a different syllabic value by context.

A Better Shorthand Analysis

Define the historical shorthand comparator independently.
Specify the unit inventory and segmentation.
Freeze the value table before lexical testing.
Replicate entropy, boundary and family diagnostics independently.
Control dictionary-collision and multiple-testing effects.
Measure coverage explicitly.
Expand unseen text and test source phonotactics.
Test Currier and scribal transfer.
Apply the same table to labels.
Preserve all failed expansions and unresolved signs.

What Would Count as a Real Shorthand Breakthrough?

Imagine an independent team reproduces the three statistical signatures under the same historical tachygraphic system.

The sign inventory is frozen.

Coverage rises substantially without adding ad hoc values.

Held-out Voynich passages expand into connected period-appropriate Latin or Italian with stable morphology.

Currier differences map to one documented shorthand variation rather than unrelated tables.

Labels become plausible names under the same values.

The historical shorthand tradition supplies independently documented forms matching the recovered functional values.

That would be transformative.

A shorthand hypothesis becomes a decipherment when the compression grammar stops finding isolated words and starts recovering connected language faster than it accumulates new exceptions.

Primary School: Write Faster by Agreeing on Shortcuts

Give “because” one invented symbol.

Give “tion” another.

A child can now compress writing dramatically.

To an outsider the page looks mysterious.

To someone who learned the convention it is ordinary writing.

Secondary School: Shorthand or Cipher?

Create two systems.

In one, symbols directly represent common syllables.

In the other, symbols deliberately disguise letters using a secret key.

Both look opaque.

The first is compression.

The second is encryption.

The exercise shows why surface mystery does not identify mechanism.

JC and Adult Readers: Compression as a Generative Channel

Formally, shorthand transforms source-language sequences into a lower-resolution observed representation.

Several source phonemes may collapse into one observed sign.

Common sequences may receive dedicated codes.

The statistical question is whether one historically constrained compression channel jointly explains the observed entropy, boundaries and word-family topology better than cipher and generator controls.

The decipherment question begins only after that channel predicts unseen connected text.

Reader Checklist: Before You Accept a Voynich Shorthand Claim

Which historical shorthand tradition is proposed?
Is it documented in the relevant period?
Are glyph correspondences functional or merely visual?
What is the basic unit: letter, syllable or word?
How much of the corpus does the fixed inventory cover?
How were candidate lexical readings protected against dictionary collisions?
Does the model explain Currier A/B?
Does it transfer across scribal hands?
Does it predict weak versus strong spaces?
Does it decode labels under the same table?
Does connected unseen text emerge?
Has an independent team replicated the results?

Frequently Asked Questions

Has Voynich been proven to be shorthand?

No. A 2026 preprint proposes a specific Italian syllabic tachygraphy model, but the manuscript remains undeciphered.

Is shorthand the same as cipher?

No. Shorthand primarily compresses conventional writing for trained users; cipher deliberately transforms information for secrecy. Historical systems can blur the boundary, but the mechanisms differ.

What did Newbold claim?

He argued that microscopic features inside Voynich glyphs encoded shorthand revealing Roger Bacon’s science. The method did not survive scholarly scrutiny.

What is new in 2026?

Matt Ruckman’s revised preprint proposes Italian syllabic tachygraphy and reports simultaneous agreement with several statistical signatures, partial lexical signals and falsifiable predictions while explicitly stating that the work is not yet a decipherment.

What is the biggest unresolved issue?

Coverage and independent validation. A shorthand system must recover connected unseen language under a stable value table rather than scattered candidate words.

Research and Further Reading

The Final Idea

Shorthand is attractive because it offers a middle path.

Not ordinary alphabet.

Not necessarily secret cipher.

A learned compression system.

That could explain why Voynich looks so structured and yet refuses ordinary reading.

The idea now has a modern quantitative proposal behind it.

That makes it worth testing seriously.

It does not make it solved.

The shorthand hypothesis will matter historically when one documented compression system can expand the Voynich Manuscript into connected language without asking the reader to supply the missing song themselves.

Voynich | Everything eduKate Knows and Tested | The Shorthand Problem: What If Voynich Is Compressed Writing Rather Than Ciphertext?