Voynich and Padua | The Expected Evidence Problem: If Padua Is True, What Should We Be Seeing That We Are Not?

A hypothesis becomes dangerous when every missing clue can be explained away.

No Paduan ownership record?

Perhaps it was lost.

No local script marker?

Perhaps the scribe was foreign.

No Paduan astronomical parameter?

Perhaps the tables were copied from elsewhere.

No local plant identification?

Perhaps the image tradition travelled.

Every one of those explanations can be historically possible.

Together, they can make a hypothesis impossible to weaken.

The Expected Evidence Problem asks a harder question: if a hypothesis is true, which observations should be common enough, robust enough or local enough that their continued absence should reduce confidence?

Absence Is Not Automatically Evidence

Most things we might wish had survived from the fifteenth century did not survive perfectly.

Archives burn.

Bindings are replaced.

Ownership marks disappear.

Leaves are lost.

Marginalia are trimmed.

Therefore:

not found ≠ never existed.

But the reverse mistake is equally serious:

possibly lost ≠ evidentially irrelevant.

Whether absence matters depends on how strongly the hypothesis predicted presence and how likely that evidence would have been to survive or be detectable.

Three Conditions for Informative Absence

The hypothesis predicts the evidence. The missing feature must be expected under the hypothesis, not invented after the search fails.
The evidence had a reasonable chance to survive or be detected. Missing ephemeral oral evidence is weak; missing repeated codicological structure may be stronger.
Rival hypotheses do not make the same prediction equally strongly. Otherwise the absence does not discriminate.

Only then does absence begin to carry weight.

What Would a Strong Paduan Production Model Predict?

Not every Paduan model makes the same predictions.

A Paduan intellectual-source model predicts less than a Paduan physical-production model.

A Paduan assembly model predicts different things again.

The Coordinate Stack therefore matters.

Paduan claim	Reasonable expected evidence
Paduan intellectual ancestry	Repeated source, image or conceptual relations unusually concentrated in Paduan networks compared with controls.
Paduan scribal training	Stable bundles of palaeographical habits matching dated regional documentary hands.
Paduan physical production	Codicological and material habits fitting a dated local corpus better than rival regions.
Paduan astronomical computation	Recoverable parameters, conventions or values that discriminate Padua from alternatives.
Paduan assembly	Quire, binding or production-unit features consistent with local workshop practice.
Paduan early ownership	Inventory, ex libris, correspondence, marginalia or another documentary trace tied to the actual object.

The stronger the claim, the more evidence families it should eventually expose.

Expected Evidence Must Be Written Before the Search

If we decide after searching what we “expected,” the test becomes too easy to manipulate.

Therefore every strong localisation hypothesis should write an expectation register first.

For example:

If C5 physical production is Padua, we expect at least some production habits to cluster with securely dated Paduan manuscripts.
If C4 scribal training is Paduan, we expect several hands or their conventions to compare favourably with local documentary controls.
If the circular diagrams are locally computed, we expect a recoverable local parameter.
If a specific Paduan image lineage is ancestral, we expect several linked rare correspondences rather than one isolated resemblance.
If early custody was Paduan, we expect a non-zero possibility of documentary intersection with relevant collections, owners or inventories.

The expectation can fail.

That is the point.

The Survival Probability Problem

Not all expected evidence is equally likely to survive.

A personal ownership note written on a lost flyleaf has low survival probability.

A ruling pattern repeated across dozens of surviving folios has much higher detectability.

A hypothetical spoken decoding instruction has almost no expected archaeological survival.

A repeated local numerical convention might survive many times if it was operationally central.

So we should distinguish:

Missing evidence	Typical absence weight
One lost ownership note	Low.
No documentary record after incomplete archive search	Low to medium.
No repeated local script bundle after broad blinded comparison	Medium to high for a strong scribal-localisation claim.
No predicted local parameter across several relevant diagrams	Potentially high for a parameter-based localisation claim.
No expected structural relation across the entire manuscript	High if the theory requires that relation.

The Missing Ownership Bridge Is Important—but Must Be Scored Correctly

We do not currently have a secure fifteenth-century ownership bridge for the Voynich.

That absence does not prove the manuscript was not in Padua.

Many books leave no surviving documentary trail.

But the absence still matters to the strength of a provenance claim.

A hypothesis with no documentary bridge is weaker than the same hypothesis with a secure bridge.

The correct treatment is not:

No record, therefore not Padua.

Nor:

Records can be lost, therefore absence means nothing.

It is:

No secure bridge currently exists; confidence must remain below the level that such a bridge would justify.

The Foreign-Scribe Escape Clause Cannot Be Unlimited

The Foreign-Scribe Problem is historically real.

A non-local-looking hand can work in Padua.

But if every hand, every ductus feature and every abbreviation habit repeatedly points away from the proposed local training environment, “foreign scribes” cannot remain a cost-free explanation.

Each escape clause should carry a burden.

Possible is not free.

The Imported-Table Escape Clause Cannot Be Unlimited Either

A Paduan astronomical manuscript can copy a table computed elsewhere.

True.

But if every recoverable parameter fits rival centres and none fits Padua, the hypothesis should lose local astronomical weight.

“Imported table” may explain one mismatch.

It should not make an unlimited series of mismatches disappear.

Expected Evidence Can Also Falsify Rival Hypotheses

This rule is not anti-Padua.

Every serious rival should face the same test.

If a German-production model predicts strong German scribal and codicological signatures, their absence matters.

If a Prague model predicts local astronomical parameters, their absence matters.

If an alchemical-function model predicts process verbs, repeated vessel roles and transformation sequences, their absence matters.

Expected evidence is a general research discipline.

The Expected-Evidence Ledger

Hypothesis claim	Predicted evidence	Detectability	Observed?	Effect if absent
Paduan physical production	Local codicological cluster	Medium–high	Open	Lower confidence if a large regional comparison fails.
Paduan astronomical parameter	Repeated Padua-compatible values	Medium	Open	Strongly lower parameter claim if relevant diagrams consistently disagree.
Paduan early custody	Documentary intersection	Low–medium	Not yet found	Keeps provenance below documentary level; not a disproof.
Single coordinated workshop	Material and production coherence	High across surviving object	Partly testable	Lower confidence if stable unit-level incompatibilities emerge.

The eduKate Expected-Evidence Test

State the hypothesis at a precise Coordinate Stack level.
Write at least three observations it predicts before searching.
Estimate whether each observation should survive or be detectable.
Search matched controls to ensure the prediction is discriminating.
Record positive findings and failures equally.
Do not replace a failed prediction with a new one without preserving the failure.
Charge every rescue explanation a complexity cost.
Lower confidence when high-detectability predicted evidence repeatedly fails to appear.
Never treat low-survival documentary absence as definitive disproof.
Publish the unresolved expectation register.

What This Adds to the Padua Programme

Padua now has a rich positive evidence environment.

The next scientific gain will come from making the hypothesis risk something.

Not just:

Here is another thing Padua can explain.

But:

If Padua is the correct coordinate for this layer, here is what we expect to find—and here is how much confidence we will lose if we do not.

Research Links

The Final Idea

A good hypothesis does not merely explain what we found.

It tells us what we should be surprised not to find.

The moment missing expected evidence is allowed to lower confidence, the Padua hypothesis becomes less comfortable—and much more scientific.