Cardan Grille Anachronism – Can a Sixteenth-Century Mask Fit the Voynich?

Here is a simple contradiction. A book made in the early 1400s appears to have been written using a tool invented in 1550. That is a 112-year gap that nobody can explain. It is a ghost in the machine; a chronological impossibility at the heart of the world’s most mysterious manuscript.

The book is the Voynich Manuscript. The theory is Gordon Rugg’s elegant hoax. The gap is the problem. Our investigation does not try to solve the manuscript itself. It follows the evidence for that single, glaring contradiction through the historical record to see where it bends, and where it breaks.

The Dating Clash We Can Measure

Two independently derived facts set the problem.

First, Accelerator Mass Spectrometry (AMS) radiocarbon tests on multiple folios date the manuscript’s parchment to the early fifteenth century. AMS measures the ratio of carbon-14 isotopes remaining in organic material and calibrates the result against long-term curves from tree-ring data. In plain terms, it pins the age of the calfskin, not the moment of writing, but it does constrain the latest plausible manufacture of the leaves.

Second, Girolamo Cardano’s grille appears in 1550. A ‘Cardan grille’ is a perforated template, typically card or parchment, that allows a writer to place a secret message into a page through cut outs, then remove the template and fill the remaining space with cover text. That is steganography, the art of hiding a message’s existence, which is different from cryptography, the art of scrambling one. The secure dates for parchment and for the grille do not overlap. The gap is at least 112 years.

These two anchors, one from physics and one from the history of ideas, are robust. Adjusting for calibration nuances cannot extend the fifteenth-century vellum forward into the 1550s. Re-dating Cardano’s published description backwards is likewise untenable within the documented literature. The contradiction sits at the centre of any claim that a table and grille procedure generated Voynich text on those leaves.

What Rugg Actually Built and Why It Matters

In 2004, Gordon Rugg demonstrated a low-technology way to generate strings that resemble Voynichese. He published the approach in Cryptologia. His version uses a prepared table of syllable-like fragments and a sliding grille to select entries in patterns that emulate observed word structures. The result is not a cipher that encodes meaningful plaintext. It is a generator of structured nonsense designed to mimic surface statistics.

The strengths of the demonstration are clear.

It shows that a human with a simple template can mass-produce words that look like the manuscript’s script without knowing any language behind it. It also forces a testable claim… if a hoaxer could do this, then one does not need to assume an unknown language or an unsolved complex cipher to explain the pages.

The weaknesses come in two forms. First, the method, as articulated, relies on a Cardan-like template, which re introduces the chronological problem. Second, critics argue that while the model hits some headline statistics, it does not capture all higher-order behaviours found by computational analyses, such as topic-dependent clustering of certain tokens and longer-range correlations across sections. That critique does not kill the hoax idea, but it raises the bar on what any generator must reproduce.

Rugg acknowledges the timing problem. He suggests two broad routes. A lost precursor to the grille existed earlier than Cardano’s publication, or the manuscript was written later, in the later sixteenth century, on stockpiled medieval vellum.

The Record Before Cardano

When we audit the literature between 1400 and 1550, we find a staircase of developments but no documented mask in the Voynich window. A useful orientation is the History of cryptology overview, which tracks the move from simple substitution to polyalphabetic practice and later steganographic devices.

• 1379: Gabriele de Lavinde compiles a papal cipher ledger. This codifies substitution alphabets, nulls, and code words, an early form of the nomenclator. No mask.
• 1467: Leon Battista Alberti circulates De Cifris and introduces the first polyalphabetic cipher with a rotating disk.
• 1499: Johannes Trithemius writes Steganographia. It outlines concealing messages within prayers and other devices that hide content without a template. Steganography appears conceptually, but still no perforated sheet.
• 1518: Trithemius’s Polygraphiae prints the tabula recta, a square table of alphabets for substitution. Tables are in play, but not masks.
• 1526: Jacopo Silvestri describes a stencil for steganography. Close to a grille and earlier than Cardano, but still too late for a manuscript written on parchment from the 1430s if writing followed manufacture.
• 1550: Cardano publishes the grille in De Subtilitate. From here the idea diffuses.

After Cardano, Giambattista della Porta’s De furtivis literarum notis consolidates techniques and describes boards with openings. In the later sixteenth and seventeenth centuries, the grille appears as a practical tool in diplomatic contexts, and variants such as trellis and rotating masks emerge. The pattern is consistent. Innovation gathered in the mid-sixteenth century onwards. The archive remains silent for the early fifteenth century.

The Grille Appears and Spreads

The invention, when it finally appears, is clearly credited. Cardano’s 1550 description of his perforated grid was treated as a novelty. Thirteen years later, in 1563, the scientist Giambattista della Porta published his own comprehensive book on cryptography, De Furtivis Literarum Notis. In it, he describes a ‘board with openings’ that is functionally identical to Cardano’s grille, suggesting the idea was already known and being adopted by other experts.

From that point on, the grille became part of the standard toolkit.

Figures like Blaise de Vigenère in France used it for diplomatic messages, and it was famously employed by Cardinal Richelieu in the 17th century. Its use after 1550 is well-documented. Its use before 1550 is non-existent.

The earliest documented precursor we have found is from a 1526 book called Opus Novum by Jacopo Silvestri, which describes a stencil for steganography. This is a significant find, noted in historical summaries of grille cryptography. It shows that the core idea of a perforated sheet did not appear completely out of nowhere with Cardano. But it does not solve the problem. A device from 1526 is still nearly a century too late to have been used on parchment made by 1438. The anachronism stands.

Could a Late Hoax Fit the Known Trail?

The first secure sighting in the manuscript’s history is a seventeenth-century letter describing Emperor Rudolf II’s acquisition, usually placed in the 1580s. That is a fertile period for cryptography, for occult interests, and for charlatans.

Rugg has named Edward Kelley as a plausible fabricator operating in this world. The proposition runs as follows… a skilled forger in the 1560s–1580s could have gained access to old vellum sheets, written out the text using a table and grille generator, illustrated in a deliberately archaic style, then sold the result into Rudolf’s court.

There are three material hurdles such a theory must clear.

1. Parchment supply: Can one obtain sheets made a century earlier in sufficient quantity and quality? Monastic and institutional stores did retain unused stock. Re-use of older leaves is known. It is not impossible that decades-old calfskin was available in the later sixteenth century.
2. Inks and pigments: Analyses report media consistent with medieval practice. A later hoaxer would either have used period-appropriate recipes or worked with materials whose composition overlaps across centuries. That is plausible, but raises the complexity of the operation.
3. Style and palaeography: The script and imagery broadly align with early fifteenth-century European hands and workshops. A forger could imitate, but sustained imitation across quires is not trivial. Minor anachronisms might be dismissed as quirks, but a close audit would look for sixteenth-century mannerisms.

The late hoax path does not contradict the chain of custody we have, because our earliest firm note already sits in the relevant century. It does, however, demand independent support from materials science or documentary finds. At present, that support is incomplete.

Language-like Behaviour Versus Hoaxable Statistics

Several studies argue that the manuscript’s text behaves like language. Certain tokens cluster by topic across sections. Some longer-range correlations appear. There are positional constraints on glyphs that feel phonotactic. Supporters of a real language or cipher take these as signs that simple random generators are not enough.

There is counter work. Algorithmic models demonstrate that many headline statistics can arise without meaning. For a detailed example of the Cardan style approach taken further, see Timm and Schinner (2019). The point is not whether a single method matches the manuscript in every respect, but whether a humanly feasible generator can imitate salient metrics. If it can, one does not need to invoke unknown language families or exotic ciphers.

This debate has a practical value for our purpose.

If higher-order behaviours turn out to be robust, repeatable and incompatible with known generators, the hoax line weakens. If, on the other hand, well-specified, hand-feasible generators can match the statistics from multiple angles, the hoax line survives. At present, neither side has carried the field.

What Would Decide the Case

The cleanest way to weigh the hoax hypothesis is to build it in public with period constraints. The plan below is designed to be auditable and practical.

Inputs

• A syllable table engineered to encode positional rules observed in the manuscript, using an alphabet size and distribution consistent with Voynichese glyphs.
• A set of fixed-size grilles and, as a control, a no-grille procedure that uses deterministic look-ups and dice rolls to replace the mask’s function.
• Period appropriate tools: quills, iron gall ink, calfskin parchment prepared to fifteenth-century recipes, and pigments documented in early fifteenth-century workshops.

Procedure

1. Generate 20 pages of text by table and grille, and 20 pages by the control method. Transcribe outputs faithfully into a page layout with the same line justification and paragraph structure as the manuscript.
2. Blind test the outputs against Voynich pages with a battery of metrics used in the literature: word length distribution, glyph positional entropy, repetition patterns, and section wise clustering.
3. Audit physical traces: stroke order, hesitation marks, error corrections, and any artefacts of copying from a generator into page layout.

Outcomes

• If both methods fail to match salient metrics, confidence in the hoax model drops.
• If either method matches well across independent metrics, confidence rises, and the role of the physical mask can be downgraded or upgraded accordingly.

This experiment does not decide the date, but it does test whether a single, disciplined scribe could plausibly have produced the text with simple apparatus. If the answer is no, we should stop pouring effort into the hoax line and concentrate on language and cipher candidates.

Forensic levers that can move the date

We will not close the 112 year gap through theory. We need friction with the object and with the archive. The following items are specific and measurable.

• Ink chronology: Micro analysis of iron gall components and degradation products to compare with dated fifteenth and sixteenth century samples. Aim to estimate the time gap between parchment preparation and writing by penetration depth and diffusion into collagen fibres.
• Binding and quire assembly: Technical study of sewing stations, thread, endbands and quire order to identify whether the present gatherings reflect a single fifteenth century assembly or a later compilation. Track re use and mis binding patterns.
• Pigment sourcing: Map pigment binders and additives against supply histories. Confirm that nothing requires a post medieval trade network. If any pigment or extender appears with a first secure use after 1500, the date moves.
• Hand comparison: Palaeographic profiling of ductus, ligatures and abbreviation marks against dated scripts. Look for unconscious sixteenth century habits in letter forms that an imitator might not suppress consistently.
• Later marginalia: Date month names, quire numbers and other Latin or European markings by hand and ink. If these can be securely tied to the later sixteenth century, they become anchors in the object’s life that favour a late hoax.

Each item is a lever that can move the hypothesis towards ‘written later’ or ‘written close to manufacture’. None requires access to hidden knowledge, only patient technical work and well chosen comparators.

Archive targets for a proto-grille

If there was a fifteenth century template or method that resembles a grille, where would we expect to see it? Three places are worth a concentrated search.

1. Cipher offices and chancery instructions: Italian city states and the papal court produced manuals and letters on cipher practice. Many remain uncatalogued or under studied. A rubric on how to structure null text or to distribute ‘cover syllables’ would be relevant even if it does not show a literal mask.
2. Technical notebooks and workshop papers: The same culture that produced cipher disks and printed tables produced practical notebooks. A sketch of a board with holes, or a grid of syllables annotated with rules, would count as a precursor.
3. Non European traditions: Claims exist of grille like ideas outside Europe. Whether those can be substantiated is a separate question. If a functional analogue predates Cardano elsewhere, it slightly reduces the prior improbability of an independent European invention.

To date, audits have found nothing dispositive. That absence cannot prove non existence, but it does preserve the weight of the anachronism.

The Assessment

On the record we have, the Cardan grille based hoax remains possible in principle and problematic in chronology.

The strongest argument for it is the ease with which a disciplined generator can hit the manuscript’s headline statistics. The strongest argument against it is the complete lack of a mask guided method in the European record before 1550, plus the need to explain higher order behaviours if those survive replication under controlled conditions.

This investigation therefore recommends two parallel tracks. First, run the replication plan with period constraints to settle whether a hand feasible generator can, in fact, match independent metrics. Second, press the forensic levers and the archive search to find anchors that move the composition date decisively towards the fifteenth or the sixteenth century. Either discovery will reduce the space for speculation.

The worst outcome is to leave the anachronism as a rhetorical puzzle. It is testable. We should test it.

Sources

Sources include: the University of Arizona’s radiocarbon dating report on the Voynich Manuscript (2009); Gordon Rugg’s foundational paper ‘An Elegant Hoax? A Possible Solution to the Voynich Manuscript’ (2004) and his subsequent research blog; primary treatises on early cryptography by Girolamo Cardano, Leon Battista Alberti, and Johannes Trithemius; modern statistical analyses of the manuscript’s text, including papers by Montemurro and Zanette (2013) and Timm and Schinner (2019); historical overviews of the field, notably David Kahn’s ‘The Codebreakers’; and the physical manuscript itself, held at Yale University’s Beinecke Library under call-number MS 408.