eduKateSG · VOYNICH · CUSTODY THROUGH TIME · MOVEMENT II
Voynich | What Can Only the Object Tell Us?
A scan can carry the visible manuscript around the world. But some questions live in thickness, pressure, chemistry, relief, sewing, translucency and material sequence. Those questions still have to return to Beinecke MS 408.
← What Can a Scan Know? · The Physical Manuscript and Its Digital Double · Voynich Research Library
There is a dangerous way to ask this question.
It is to imagine that “the object” is mystical: that the physical Voynich Manuscript contains some privileged aura that a digital image cannot possess, and that serious research therefore becomes more authentic merely by touching the parchment.
That is not the argument.
The scientific distinction is much narrower.
Some variables are not adequately encoded in the available representation.
If the research question depends on one of those variables, the question must return to the material source—or to an instrument that measures the source directly.
That last phrase matters. “Object-only” does not mean the knowledge must remain locked inside a reading room forever. A microscope can create photomicrographs. Spectroscopy can create spectra. X-ray methods can create elemental or structural measurements. Multispectral imaging can expose signals absent from ordinary RGB photography. Three-dimensional capture can turn surface relief into a digital field.
So the better term is:
OBJECT-DEPENDENT AT ACQUISITION
The physical manuscript is required to acquire the missing signal. Once acquired responsibly, that signal can become a new representation and travel through the research world without requiring every later scholar to touch the codex again.
This is the mirror image of the previous article. What Can a Scan Know? asked how far digital evidence may legitimately travel. This article asks when the scan has reached its material ceiling—and what scientific job would justify asking the original manuscript to supply another channel.
Yale’s current policy makes this distinction operational: original-object access is restricted because of fragility and demand, and researchers must explain why the original rather than scans and existing scholarship is necessary. See Yale Beinecke — The Beinecke Cipher (Voynich) Manuscript.
The Object Is Not One Extra Data Point
A flat visible-light image compresses many physical dimensions into a small number of recorded channels.
The material manuscript contains more variables than RGB intensity at x and y coordinates.
- surface height;
- surface roughness;
- sheet thickness;
- local stiffness;
- translucency;
- fibre and skin structure;
- hair-side and flesh-side characteristics;
- chemical composition;
- microscopic particle morphology;
- ink penetration;
- layer order;
- pressure and indentation;
- cut geometry;
- sewing and binding mechanics;
- edge wear;
- fold memory;
- contact transfer;
- subsurface traces;
- three-dimensional deformation.
Some of these variables leave visible consequences in ordinary scans. Some can be partially inferred. Some require specialised imaging. Some require microscopy, spectroscopy or sampling.
The object is therefore not “the scan, but better.”
The object is the source from which additional measurement channels may still be acquired.
That is why access to the original can matter even when the public digital images are excellent.
1. Chemistry: A Colour Is Not a Material
The clearest example is chemical composition.
A scan can tell us that two regions appear blue.
It cannot, from visible appearance alone, establish that both contain the same pigment.
Different materials can produce similar visible colours. The same material can look different under different illumination, concentration, ageing, substrate and processing conditions.
The Voynich Manuscript has already demonstrated why object-derived chemical evidence matters. In the 2009 McCrone Associates analysis linked by Yale, tiny samples were collected using a fine tungsten needle and microscalpel. Portions were analysed using polarized light microscopy, infrared spectroscopy, micro X-ray diffraction and energy-dispersive X-ray spectrometry in a scanning electron microscope.
Those methods identified material constituents in sampled inks and pigments—including iron-gall inks and mineral or copper-based pigments—that a normal colour image cannot establish chemically by appearance alone.
The report also contains an important limitation that should become a general rule: the results apply to the samples analysed.
Material identification at sampled loci ≠ universal chemical identity across every visually similar locus.
This is object science at its best. The object is asked a material question. A minimally sized sample or non-destructive instrument acquires a new signal. The result then becomes shareable evidence—but only with its sampling coordinates and method attached.
Primary analysis: McCrone Associates — Materials Analysis of the Voynich Manuscript. Yale links this report from the manuscript’s official collection page.
2. Surface Relief and Pressure: The Page Has a Third Dimension
Writing is a physical act.
A pen does not merely deposit a two-dimensional colour.
It touches a surface.
Pressure, direction, nib geometry, speed, parchment response and later flattening can leave material traces.
Ordinary photography may show some consequences: darker starts, tapered ends, pooling ink, apparent stroke overlaps, ridges or depressions under raking light.
But a conventional front-lit scan does not encode the full height field of the page.
visible stroke shape ≠ complete stroke event
This matters when researchers ask questions such as:
- Was a faint line drawn or merely stained?
- Did a pen physically indent the parchment?
- Does a mark cross another mark or only appear to?
- Was a line ruled before writing?
- Can pressure patterns distinguish a deliberate writing stroke from later contact damage?
- Are apparently separate components physically continuous?
These may require microscopy, raking-light imaging, reflectance transformation, profilometry or another method that acquires surface geometry from the object.
Once acquired, the resulting relief map can travel digitally.
But the data did not exist in an ordinary RGB scan merely because that scan had many pixels.
3. Layering and Deposition: Which Mark Came First?
One of the most powerful object-level questions is chronological at microscopic scale.
At a crossing:
A over B or B over A?
A public image may make one answer look obvious.
It may be wrong.
Ink absorption, pigment opacity, ageing, abrasion and image processing can confuse visual overlap. Microscopy can sometimes resolve which deposit physically overlies another. Spectral or chemical methods can distinguish two materials that look similar in ordinary light.
These micro-sequences matter because they can constrain production order:
- text before drawing or drawing before text;
- outline before pigment or pigment before later outline;
- original mark before later correction;
- ruling before writing;
- folio number after earlier text;
- repair after damage;
- later annotation above the original surface state.
No single overlap decides the history of the manuscript.
But a repeated pattern of object-confirmed micro-sequences can reconstruct production stages that a flat scan can only suggest.
Chronology can hide in thickness measured in microns.
4. Parchment Is Skin, Not Background
Digital analysis often treats parchment as the pale rectangle behind the interesting content.
Codicology treats the parchment as evidence.
Animal skin has direction, surface character, thickness, follicles, scars, holes, edge shape and production traces. Preparation stretches and scrapes the material. Hair and flesh sides can behave differently. Defects can be stitched or worked around. A scribe chooses where to place writing relative to those features.
Many large parchment features are visible in good scans.
The finer ones may not be.
This distinction already has its own specialist room in the library: The Parchment-Side Problem: Hair, Flesh and What the Skin Can Still Tell Us.
The methodological point here is broader.
Parchment carries production evidence that is independent of the script’s meaning.
A material localising hypothesis might eventually depend on preparation practice, sheet selection, side patterning, thickness distribution or another codicological feature.
That kind of claim cannot be established by identifying a plant drawing or finding a statistical text cluster.
It has to be earned in the material lane.
5. Thickness, Flex and Fold Memory
A page has mechanical properties.
Thickness affects folding.
Stiffness affects how a leaf opens.
Repeated folding can create mechanical memory.
Cockling and deformation can record moisture history, tension or later storage conditions.
A scan can show the outline of a fold.
It does not automatically tell us the force required to open it, the thickness of the sheet, the three-dimensional ridge profile or the relationship between stiffness and current binding.
For the Voynich foldouts this matters especially.
The physical interface has states:
closed → partly unfolded → fully unfolded
A flat digital image usually gives us the final state beautifully.
The object can preserve evidence about the path between states.
That path may matter if the manuscript was designed for sequential revelation, portable use, repeated folding or another interface behaviour.
The object may therefore answer not only what is on the page? but what physical actions were possible or habitual?
6. Sewing and Quire Mechanics: A Book Is a Machine
Page images encourage us to think in folios.
The physical codex was made from gatherings.
Bifolia were folded, nested, sewn, bound, repaired, numbered and sometimes rearranged. Missing leaves can leave stubs. Conjugate relationships constrain what once belonged together. Sewing stations and attachment structures can preserve evidence of binding history.
Some of this is reconstructed very successfully from images and cataloguing.
Some questions depend on physical relationships that a sequence of isolated recto-verso scans abstracts away.
The specialist library already treats this in The Binding Problem, Missing Leaves and Quire Reconstruction and The Bifolium as a Production Unit.
The methodological rule is simple:
Before reading a sequence semantically, establish which parts of the sequence are physically historical and which are later interface decisions.
The object can constrain original order even when it cannot translate one line.
7. Translucency and Show-Through: Light Direction Changes the Evidence
Parchment is not perfectly opaque.
Ink or pigment on the opposite side can influence what is visible. Surface stains can be confused with show-through. A faint feature may appear or disappear depending on whether the page is front-lit, back-lit or photographed under another illumination geometry.
This matters for any claim involving:
- faint hidden marks;
- possible erased text;
- contact transfer;
- apparent underdrawing;
- alignment between recto and verso;
- ink penetration;
- features seen only near thin parchment areas.
The right test may not require a researcher to touch the manuscript directly.
It may require the institution to acquire a different image under transmitted light or another spectral condition.
OBJECT-dependent does not imply HAND-dependent.
This is an important preservation principle. If an instrument can recover the required material signal without unnecessary handling or sampling, the research question can be escalated to the object while still minimising risk.
8. Contact Transfer, Wear and Dirt: Use Leaves Traces
A manuscript remembers handling imperfectly.
Facing pages can transfer material.
Edges can darken.
Repeated opening can stress particular folds.
Areas touched often can wear differently from protected areas.
Damage and repair can create their own timeline.
Many of these effects are visible enough to study from high-resolution imagery. But object-level examination can help discriminate among alternatives that look similar in a flat photograph: transferred pigment versus staining; abrasion versus faint writing; cut versus tear; original hole versus later damage; contact mark versus show-through.
The existing specialist articles carry those details: The Contact-Transfer Problem and The Damage and Repair Problem.
Here the larger lesson is:
Use history can be material even when the manuscript’s semantics remain unknown.
9. Subsurface and Erased Evidence: What Ordinary Vision Cannot Exclude
An ordinary scan can show no earlier writing.
That is not logically identical to proving that no earlier writing ever existed.
The relevant scientific question becomes:
Which acquisition channel would be expected to reveal the proposed trace if it were present?
Microscopy, multispectral imaging, ultraviolet or infrared response, transmitted light and other non-destructive methods can sometimes test faint, erased or subsurface features more strongly than ordinary visible-light imaging.
The eduKate library treats the question directly in The Palimpsest Problem: Was Anything Written Under Voynich?.
The methodological principle is more general.
Negative evidence is only as strong as the sensitivity of the channel used to look for it.
absence under RGB ≠ absence under all material channels
And even a specialised negative result should remain tied to the method’s detection limits.
10. Sampling: The Object Can Answer Questions by Losing Material
This is the most ethically difficult category.
Some measurements require physical material.
The sample may be tiny.
It may still be irreversible.
The McCrone analysis is a useful example. Material was physically collected from specified loci and distributed among several analytical methods. The knowledge gained was significant precisely because the analysis did not rely only on appearance.
But the sample cannot be unremoved.
Radiocarbon dating creates an even clearer conservation equation: material from parchment can constrain when the animal lived, but consuming sample material reduces what remains for future analyses.
That gives destructive or micro-destructive testing a much higher threshold than ordinary scan analysis.
Expected information gain must justify irreversible evidence consumption.
And the calculation must include future optionality.
A method invented twenty years from now may extract more information from less material than a method available today.
So “we can measure this” is not yet equivalent to “we should sample this now.”
The Object-Dependent Evidence Matrix
| Question | Scan contribution | Object-dependent acquisition | Likely class |
|---|---|---|---|
| Where is a visible glyph? | Usually decisive | Normally unnecessary | SCAN |
| Is a faint line physically indented? | May generate candidate | Relief / raking-light / microscopy may be needed | BOTH |
| Are two blue regions chemically the same? | Appearance only | Spectroscopy / material analysis | OBJECT / BOTH |
| Which of two crossing marks lies physically above? | May suggest | Microscopic deposition evidence may decide | BOTH |
| How thick is this parchment leaf? | Not reliably encoded by ordinary scan | Direct or non-contact thickness measurement | OBJECT |
| Does a foldout contain a repeated visible motif? | Usually decisive | Not normally required | SCAN |
| How does the fold physically open and resist? | Partial | Object / physical facsimile / mechanical observation | BOTH |
| Is a stain actually transferred pigment? | Candidate evidence | Microscopy / chemistry may discriminate | BOTH |
| Is an apparent missing feature truly absent materially? | Depends on detection channel | Specialised acquisition may be required | BOTH / UNRESOLVED |
| What elements or compounds occur in a sampled ink? | Cannot establish composition from appearance | Instrumental material analysis | OBJECT |
The purpose of the matrix is not to make object access sound exotic.
It is to stop us asking the wrong evidence channel to do the wrong scientific job.
The Most Important Refinement: Object-Only Can Become Scan-Available Later
This is where the four-class system becomes dynamic.
Suppose a question is OBJECT today because ordinary images do not encode surface relief.
A future institution performs a high-resolution three-dimensional capture of the relevant folios.
The resulting dataset is published with calibrated metadata.
Now many later researchers can test the relief question without touching the original.
OBJECT-dependent acquisition → new representation → later SCAN/data-testable research
The same can happen with spectral cubes, micrographs, chemical spectra, radiographic images and other instrument outputs.
This means the purpose of object access should often be to create a reusable high-fidelity measurement rather than to give one researcher a private encounter with the manuscript.
That maximises the return on the preservation cost.
If the object must be asked, ask it in a way that lets the answer travel.
The Object Access Threshold
Yale requires researchers seeking the original to explain why access to the original is necessary and to demonstrate the scholarly contribution of the proposed work.
The Voynich method can make that threshold even more explicit.
- Define the unresolved hypothesis.
- Name the decisive material variable.
- Show that the existing representation does not encode it adequately.
- Identify the least invasive acquisition method that could recover it.
- Predeclare the outcomes that would support, weaken or fail to discriminate among hypotheses.
- Estimate handling or sampling cost.
- Explain why the expected information gain exceeds that cost.
- Design the acquisition so the resulting data can be preserved and reused.
- Version the resulting representation and attach it to the exact folio / locus.
- Return uncertainty with the measurement.
This turns access into an experiment rather than an experience.
The original is not where a hypothesis goes to feel important. It is where a hypothesis goes when the missing variable actually lives there.
Padua: The Object Lane Could Matter Enormously—and Still Prove Nothing
The Padua hypothesis makes object-dependent evidence especially tempting.
Perhaps parchment preparation will localise production.
Perhaps an ink or pigment will reveal a regional practice.
Perhaps ruling, sewing, quire assembly or microscopic sequence will match a northern Italian comparator.
Those are valid possibilities.
They are not automatic wins.
A material feature becomes localising only if it is demonstrably discriminating.
object evidence ≠ geographic evidence
geographic evidence ≠ Padua evidence
An iron-gall ink may establish material consistency with a historical manuscript tradition while remaining geographically widespread.
A parchment preparation feature may occur across multiple regions.
A sewing structure may be altered by rebinding.
A pigment may have circulated through trade.
So any Padua object test needs matched material controls.
The correct question is not:
Can the object produce a Paduan-looking signal?
It is:
Can the object produce a signal that Padua explains better than Bologna, Venice, Vienna, southern Germany, Bohemia or another plausible production world?
Object access raises the quality of the evidence channel.
It does not remove the need for controls.
What the Object Still Cannot Tell Us Directly
Physical access is powerful enough that it creates another form of overconfidence.
The object can tell us material facts.
It does not automatically tell us semantic facts.
- A pigment composition does not translate a label.
- A parchment date does not name the scribe.
- A sewing sequence does not identify the language.
- A microscopic overlap does not establish the manuscript’s purpose.
- A hair-side pattern does not prove a city without comparative localisation evidence.
- An iron-gall ink does not tell us what the text says.
- A pressure pattern may identify a writing event without identifying its author.
- A material correction can establish sequence without explaining intention.
This is the object version of the same discipline we imposed on scans.
Better evidence channel ≠ unlimited claim scope.
The object tells us what happened materially.
Meaning still has to be built through independent bridges.
The Object Is a Future Instrument Too
Every generation risks imagining that its current instruments are the final instruments.
They are not.
Future imaging may recover surface topology at finer resolution without contact.
Future spectroscopy may separate materials more precisely.
Future algorithms may integrate spectral, geometric and chemical channels into representations we do not yet possess.
Future sampling methods may require less material.
Future comparative databases may make a currently non-localising material feature suddenly informative.
This is why preserving the physical object preserves more than the evidence we currently know how to read.
The original stores future measurement channels that have not yet been invented.
That may be the strongest reason for restraint.
We do not know what the year 2126 will be able to ask of the same parchment.
So preservation keeps an option alive for a receiver who does not yet exist.
The Object Contract
The mirror of the Scan Contract is now ready.
Before an object-level Voynich claim is promoted, record:
- Exact object coordinate: folio, recto/verso, locus and physical structure.
- Research question: what uncertainty is being reduced?
- Material variable: chemistry, relief, thickness, layering, stitching, translucency, wear or another physical property.
- Acquisition method: visual examination, microscopy, spectral imaging, X-ray method, sampling or other technique.
- Invasiveness: non-contact, contact, micro-destructive or destructive.
- Detection limits: what the instrument could miss.
- Calibration / controls: how instrument and comparator behaviour are established.
- Transformation path: how object signal becomes image, spectrum, measurement or dataset.
- Claim boundary: what the result establishes and what remains inferred.
- Future receiver packet: data, metadata, method, uncertainty and enough lineage to reproduce or reinterpret the result later.
The object contract protects both sides.
It protects the manuscript from unnecessary intervention.
And it protects the research from the assumption that physical access automatically turns speculation into fact.
The Full Escalation Ladder
The scan and object articles now combine into one research route:
- SCAN. Test the claim against the highest-quality available representation appropriate to the question.
- REPRESENTATION ROBUSTNESS. Check alternate scans, transcription regimes, crops, processing settings or representations.
- NAME THE CEILING. Identify exactly which decisive variable remains unencoded.
- ALTERNATE NON-INVASIVE CHANNEL. Ask whether better imaging or an existing object-derived dataset can recover it.
- OBJECT PROPOSAL. If necessary, define an object-dependent acquisition with explicit information value.
- LEAST-INVASIVE MEASUREMENT. Acquire only what is needed.
- DISTRIBUTE THE RESULT. Convert the measurement into a reusable, versioned representation.
- RETURN TO CONTROLS. Test whether the new object signal actually discriminates among competing hypotheses.
- PRESERVE NULL. If the result fails to discriminate, do not force a conclusion.
This makes the relationship between physical and digital research cooperative rather than adversarial.
SCAN finds where the question lives. OBJECT supplies channels the scan lacks. REPRESENTATION carries the new answer back out into the world.
World Return: The Object Is Where the Missing Channels Wait
The physical Voynich Manuscript is not important because physical things are somehow more intellectually noble than digital things.
It is important because it remains the highest-order surviving source.
The scan can carry visible geometry.
The transcription can carry symbolic sequence.
The dataset can carry measurements.
The paper can carry interpretations.
But when a question depends on chemistry, thickness, relief, deposition, sewing, translucency or another unrecorded material variable, all of those representations eventually point backward.
Back to Beinecke MS 408.
Not because the object should be handled casually.
Precisely because it should not.
The original stores signals that have not all been measured.
It stores signals that may not yet have names.
It stores future experiments.
So a responsible present generation should neither worship the object nor exhaust it.
We should use the digital representation aggressively where it is the right instrument.
We should escalate to the object precisely when the missing variable requires it.
And when we do ask the object a question, we should turn the answer into a durable representation so the next thousand researchers do not have to ask the parchment the same question again.
The best object-level experiment does not end at the object. It returns a new, traceable signal to the informational river.
That is how physical custody and epistemic custody begin to cooperate.
Next: Voynich | The Representation Has a Material Ceiling
Primary institutional routes: Yale Beinecke — Voynich Manuscript access, physical description and original-object policy · McCrone Associates — Materials Analysis of the Voynich Manuscript.
What Can a Scan Know? → · When the Object Stayed and the Information Escaped → · Custody Through Time → · Voynich Research Library →