Research ProgrammesMethodology · RAI
Relational Archival Investigation
Follow the evidence.
Test the connection.
A source-led protocol for discovering and testing historical problems through relations among documentary assertions.
The evidential path
Choose a section. Your place in the investigation is kept as you move between them.
Definition
A method that keeps the question
and its evidence together.
Relational Archival Investigation is a source-led historical research protocol in which relationships among documentary assertions generate or revise candidate research problems; the relation, its explanation, and its consequences are adjudicated separately; and the evidential path is preserved as a versioned research output.
RAI begins with a field of inquiry and allows relations among records to change the question. It does not presume an archive without interpretation: identifying a record, translating it and connecting it to another record are themselves acts that require reasons.
Relation-led problem formation
Let a comparison generate or revise the question. Record why the connection became a historical problem, including a premise that the evidence defeats.
Claim-specific verification
Ask what a particular source can establish about a particular claim. Identity, date roles, translation and source dependence need their own checks.
Separate adjudication
Establishing a documentary relation does not establish its cause or its effect. Assess the relation, its explanation and its consequences separately.
Preserve the evidential path
Keep the sources, assertions, warrants, alternatives and decisions with the result, so another reader can inspect and reopen the inquiry.
Five distinctions the method retains
- Referent
- The person, work, object, event or activity under investigation.
- Witness
- The particular record through which information is available, including its consulted version and access level.
- Assertion
- What a passage, field, inscription or image is reported to say or show.
- Proposed relation
- The connection being tested between assertions or their referents.
- Historical claim
- What the investigator concludes from the relation, with its grounds and limits.
The eight stages of inquiry
Procedure
Eight stages. Not a one-way journey.
These are ordered obligations, not a once-only sequence. A failed match can reopen selection; a new annotation can change the question. Record iterations that change the inference.
Stage 01 / 08
Define the field of inquiry
What is the field of inquiry—and what lies outside it?
Specify the collection, series, publications, performances or objects to be examined. State the initial interest, languages, period and access limits. Keep exploratory additions distinct from the corpus supporting a result.
- Keep in the record
- Corpus boundary, selection rationale, consequential exclusions and the corpus state supporting the result.
- Boundary
- An instructive selection of cases is not a sample from which to estimate institutional error rates.
Stage 02 / 08
Record the sources and their statements
What was actually inspected, and what does it state?
Give each consulted witness a stable local identifier and precise locator. Preserve original wording separately from translation and normalized search terms. Distinguish the date of a record from dates assigned to its subject.
- Keep in the record
- Source identifier, locator, access level, relevant assertion, language, date roles and known version information.
- Boundary
- A catalogue description, a transcription and an inspected image are different levels of access.
Stage 03 / 08
Establish the grounds of comparison
Same in what respect, and on which grounds?
State exactly what is being linked: the same person, role, design, impression, work lineage, material realization or performance. Explain why the matching grounds fit that scale. Preserve meaningful differences.
- Keep in the record
- Candidate-link statement, scale of identity, matching grounds and remaining identity alternatives.
- Boundary
- A shared title or year can locate candidates without establishing identity.
Stage 04 / 08
Distinguish the claims being made
Which assertions are being joined, and what is the join expected to establish?
Separate components that depend on different evidence. A named performer in a dated photograph may involve distinct person, role, event and image-production claims. Make only the distinctions the question requires.
- Keep in the record
- Constituent assertions, the relation under examination and the proposition it is expected to support.
- Boundary
- A date difference and a documented amendment history are not the same relation.
Stage 05 / 08
Frame the question and consider alternatives
What would make us change the question or reject its premise?
State what remains unsettled, why it matters and what evidence could distinguish plausible explanations. Include an account under which the source is responsible and the initial suspicion is mistaken. Credit prior scholarship.
- Keep in the record
- Problem memorandum, prior accounts, competing explanations and discriminating evidence required.
- Boundary
- Entering an institution’s existing explanation in a new register does not make it a new discovery.
Stage 06 / 08
Examine corroboration and dependence
What support survives when an uncertain link is set aside?
Return to the source level on which the conclusion depends. Check whether agreeing descriptions share an attribution. Ask what remains established if a disputed connection is removed.
- Keep in the record
- Tests completed, sources used, dependence status, alternatives retained or rejected, and surviving conclusions.
- Boundary
- Different repositories do not automatically provide independent witnesses.
Stage 07 / 08
Assess explanations and historical consequences
What is established, what explains it, and what followed from it?
After establishing the relation, investigate its explanation. A claim about its effect on a later essay, exhibition or institutional decision needs another evidential connection. Distinguish possible effects from demonstrated uptake.
- Keep in the record
- Separate dispositions for the documentary relation, its explanation and any asserted historical or practical consequence.
- Boundary
- A later description can be better supported without revealing the process that produced it.
Stage 08 / 08
State the findings and grounds for reconsideration
What can another researcher retain, challenge or revise?
State the strongest warranted conclusion. Preserve essential rejected matches and resolved apparent conflicts. Identify the specific evidence that would reopen a supported, qualified or unresolved claim.
- Keep in the record
- Conclusion, claim-level status, limitations, source and decision identifiers, and the reason for closure or reopening.
- Boundary
- Do not upgrade every conclusion because new evidence settles only one part of the case.
The procedure scales to the question. A compact dossier can be sufficient; more fields are not automatically better research.
Sources, statements, connections and decisions
Research record
Four registers.
One inspectable path.
A claim-bearing relation keeps the source assertions, grounds of comparison, warrant, alternatives, disposition and reopening condition together. It can record a rejected link as clearly as an accepted one.
Source register
The witness and its consulted state.
Creator or issuer; precise locator; access level; language; version; consultation date.
Assertion register
What the source is reported to state.
Original wording or attributed description; translation; date role; source reference.
Relation register
The connection, its question and its warrant.
Identity scale; grounds; alternatives; source dependence; discriminating test.
Decision register
What was accepted, qualified, rejected or left open.
Reason; claim-level status; responsibility; version; limitation; reopening condition.
The source and the decision remain distinct
Revising an interpretation must not overwrite the source assertion on which it was based.
“Not examined,” “not accessible,” “not located,” “examined without finding,” and “demonstrated absent from a defined inventory” are different evidence states. A missing search result does not establish historical absence.
Intellectual context, PIC and the limits of the method
Intellectual context
Founded through PIC.
Governed by the evidence.
Post-Interpretive Criticism supplies the intellectual departure: explanations acquire authority and must remain answerable for what they claim. RAI translates that concern into documentary and inferential obligations.
A successful investigation need not confirm PIC. An institutional explanation may protect a historical distinction, and responsibility may require more explanation rather than less.
Museum of One does not require its investigations to confirm the theory from which the institution emerged.
A specified synthesis, not an invented history of method.
RAI draws on historical source criticism, Collingwood’s relation between question and evidence, Ginzburg’s clue-based inquiry, Toulmin’s distinction between grounds and warrants, Wigmorean evidence analysis, abduction, and assertion-centred and provenance models.
Its proposed contribution is an archival-historical protocol that preserves the history of a question alongside the support, scope and revision of its answer. It does not claim to originate comparison or a general logic of discovery.
Scope, limitations and evaluation
Worked demonstrations establish how a protocol can be used; they do not establish universal priority, superior discovery, independent validation or the prevalence of errors across institutions.
The evaluation proposed in the formulation separates historical quality, traceability and revisability, and efficiency. Comparison must be with competent existing practice under comparable access and expertise—not with an artificially weak alternative.
Language, material knowledge, access limits and tacit expertise remain consequential. The method needs revision if its records burden inquiry without improving inspection, encourage trivial anomaly collection or misrepresent local evidence.
Rights and permissions remain separate from scholarly annotation. This page does not distribute institutional images, private correspondence or unpublished manuscripts.
RAI · An illustrated investigation
An argument,
made inspectable.
Follow the evidence. Test the connection. See what can be claimed.
Three constructed investigations make the method visible. Nothing here is an unpublished case, an archival reproduction or an automated historical judgment.
Browse the evidence archiveConstructed teaching example
Stated support Withheld or not established Every line has an inspectable basis.
Findings and alternative explanations
What changed?
No assumption has been changed yet.
What remains?
The recorded assertions remain inspectable.
What is needed next?
Why not conclude:
Sources, passages and research decisions
The research record
Keep the route, not just the result.
Read this example without the diagram
Changes made while exploring are temporary and stay in this page. They never amend a source or a published research decision.
The examples are fully readable below. With JavaScript available, this section also becomes an interactive evidence workspace.
A difference — Two dates. Is one of them wrong?
All records, identifiers, names and dates in this exercise are constructed.
Question
Do these descriptions assign different production years to the same work?
Acquisition list
Production year: 1912 — Production-year assignment.
Names Study in indigo, Maker A, oil on board, 40 × 55 cm. These matching fields support a work-level comparison in this exercise; they are not a physical examination. A date assigned in a record does not independently authenticate when an object was made.
Collection description
Production year: 1914 — Production-year assignment.
Uses the same maker, title, medium and dimensions. Both leading years explicitly describe production assignments, not an exhibition or a later print. No amendment log or administrative rationale is present in this baseline record.
Proposed connection
Maker, title, medium and dimensions correspond. The identity scale is the described work, not every physical component.
Warrant
If the descriptions concern the same work and the date fields have the same role, their unequal years establish a difference in published assignments.
Alternatives
Different works; a differing date convention; an earlier error; or additional evidence available to a later compiler.
What the comparison teaches
Keep a supported documentary difference separate from an unproven explanation.
A dated decision record tied to this work could explain the difference. Independent historical evidence would be needed to establish the actual production year.
Reopening condition
An accession crosswalk, an image that defeats the match, a clear column convention, or a dated decision record.
A justified difference — Different dates. Different occasions.
All witnesses and dates are illustrative. This is not a summary of an unpublished investigation.
Question
Are these dates competing assignments, or dates of different events?
Exposure register
Image exposure: 1921 — Photographic exposure.
Records an exposure with identifier N-07. It identifies the image, but says nothing about when a later physical print was produced. An exposure date is not a printing date or a date of later exhibition.
Print record
Printed in 1984, from N-07 — Material printing.
Explicitly attributes the print to negative N-07 and states 1984 as its production date. This is a later realization of the image. The physical print need not date to the original photographic exposure.
Proposed connection
The print record explicitly identifies negative N-07. The link concerns image lineage, not an unchanged physical object.
Warrant
An image made on one occasion can be printed on another. Different event roles make these date assertions compatible.
Alternatives
A copied or incorrect negative reference; a misread date role; a later reproduction or a distinct image.
What the comparison teaches
A resolved apparent conflict can change the question rather than end the investigation.
Production and exhibition documentation could explain how this later print was made and used; the initial date conflict does not survive the date-role check.
Reopening condition
Original negative evidence, a production file, or documentation revising the print’s source identification.
Agreement & dependence — Three records. How many origins?
This is a constructed provenance exercise, not a claim about any real archive.
Question
What does each agreeing description independently contribute to the attribution?
Attribution register
Attributed to Maker A — Attribution, not a date.
The account supplies an attribution. Its own evidential grounds have not been independently authenticated in this example. A source for an attribution is not automatically proof that the attribution is correct.
Digital catalogue
Attributed to Maker A — Attribution, not a date.
Repeats the attribution. Selecting ‘Trace dependence’ reveals the exercise’s stated derivation from the attribution register. A different repository address does not prove a distinct observational origin.
Exhibition guide
Attributed to Maker A — Attribution, not a date.
Repeats the same attribution. In the traced state, it too cites the earlier attribution register. Repetition can document circulation without adding independent corroboration.
Proposed connection
In the traced state, the catalogue and guide expressly cite the earlier register for this attribution.
Warrant
Descriptions derived from the same attribution do not become independent observations merely by appearing separately.
Alternatives
Separate examination; a shared source; an undocumented mixture of origins. An unknown ancestry stays unknown.
What the comparison teaches
Count the support a source contributes, not simply the number of places a claim appears.
Inspect the grounds of the original attribution and seek a genuinely independent observation. Keep untraced ancestry explicitly unknown.
Reopening condition
A documented independent examination, a source citation showing different ancestry, or a revised attribution.
The Evidence Room · Research collections
The Evidence Room
An archive of research materials: notebooks, photographs, catalogues, historical records and research notes. Browse the documents, consult their provenance, and follow the investigations to which they contribute.
A researcher’s note documents an investigation. It is not automatically an independent witness to the historical event. Handwriting in an archive retains its original creator and provenance.
Illustrative catalogue. These typeset examples demonstrate the reading room. They are not archival reproductions, your handwritten notes, or released studies.
Public collection
Space for the evidence.
Care for its context.
No archival material has been released here yet. Selected records will appear with their creator, collection, dates, source references and conditions of use.
The illustrative catalogue lets you try browsing and close reading without publishing any of your research material.
No records match these selections. Change a filter or clear your search.
The authentic evidence register
Prepared for inspection.
Released with care.
No authentic dossier has been released to this page yet. Only individually selected, attributed and cleared material will appear here—not entire notebooks or unpublished study files.
The viewer demonstration is typeset and explicitly illustrative. It is not a facsimile of archival handwriting.
About these collections: sources, notes and interpretation
- Archival witness
- The creator, archive, collection, original identifier, exact locator, date roles, consulted version and access level. A facsimile, catalogue description and transcription are not interchangeable.
- Research observation
- The researcher’s identity, date and version, with original observation separate from transcription, translation and later interpretation. Uncertain readings remain marked.
- Claim-bearing relation
- The matching grounds, scale of identification, warrant, alternatives, dependence and evidence that would reopen the connection.
- Research finding
- Separately assessed documentary, relational, explanatory and consequence claims. Earlier decisions remain visible when later readings change.
Some records can be described but not reproduced. In those cases, the dossier provides precise locators and appropriate source links rather than exposing protected images or correspondence.
Public release and research completeness are separate decisions. An evidence dossier is a research record, not institutional endorsement, peer review or a claim to ownership of the source.
Applications and related work
Follow the method into the research.
The application register links only to explicitly released study overviews. A public overview is not a release of its manuscript, and a link is not evidence of independent validation.
No public application record is displayed here yet. Unreleased investigations remain outside this register.
Publication and citation
The method, its evidence,
and its demonstrations.
Three independently citable publications form the first release. Read the foundational paper alongside its documentary supplement and extended case studies; each retains its own DOI, edition and rights record.
- Formal name
- Relational Archival Investigation (RAI)
- Formulated by
- Dorian Vale
- Developing institution
- Museum of One
- First manuscript formulation
- Public edition of this overview
- V.1 · First edition
- Original paper publication date
- This web overview first published
The original publication date follows the Zenodo records. It is separate from the manuscript’s revision date and this web page’s publication date.
Foundational paper · Version 1.0
Relational Archival Investigation: Identity, Chronology, and Historical Judgment in Korean Experimental Art
The protocol, its intellectual context and the historical questions it addresses.
Dorian Vale · Museum of One · · In Copyright
Supplement 1 · Version 1.0
Documentary Evidence and Source Comparisons: Supplement 1 to Relational Archival Investigation
Documentary evidence and source comparisons accompanying the main paper.
Dorian Vale · Museum of One · · In Copyright
Supplement 2 · Version 1.0
Extended Case Studies and Methodological Demonstrations: Supplement 2 to Relational Archival Investigation
Extended cases showing how relations, explanations and consequences are assessed separately.
Dorian Vale · Museum of One · · In Copyright
From the publication record
What changes when the records are compared?
Accurate statements can produce an inaccurate history when they refer to different objects or occasions. RAI examines the grounds for joining them. The following examples summarize findings already published in the main paper and its supplements. They are distinct from the three constructed teaching exercises in the interactive explorer. These published demonstrations do not establish independent validation of the method.
Lee Kang-so (이강소): production and presentation
For Painting–78–1, MMCA’s acquisition entry 75 assigns 1978 while collection record 7924 gives 1977. A published conversation dated 14 October 2018 records Lee Kang-so distinguishing production in 1977 from presentation in 1978. The comparison therefore supports an attributed production–presentation sequence. It does not establish why the annual-report compiler used 1978, when a museum field changed, or why the two records give different durations. The methodological result is to preserve the supported chronology while keeping those separate questions open.
Main paper, PDF pp. 7–8 · Supplement 1, PDF p. 4, C-75.
Park Suk-won (박석원): a supported difference and a rejected identification
For Jeok–Eungnyeok, acquisition entry 76 gives 1974 and collection record 7925 gives 1972, alongside a separate statement that the wooden elements were remade in 2016. Distinguishing the remaking leaves the production-year difference intact. The comparison does not establish which year is correct or why the assignments differ. A related, separately numbered work dated 1973 has different materials and dimensions; it does not supply a third date for this installation. RAI retains the unresolved comparison and the rejected extension as different judgments. Accession documentation, a contemporary photograph or an explicit connection between the works could reopen a particular claim.
Main paper, PDF p. 9 · Supplement 1, PDF pp. 4 and 6–7, C-76 and section C.
Kim Youngjin (김영진): one work, different presentations
For 1978–2, catalogued by MMCA under KIM Youngjin b, both leading years are 1978. Collection record 7928 distinguishes an early presentation on one monitor from the twelve-monitor arrangement at the 2016 Busan Biennale. The work’s historical year and a later component count cannot together establish that twelve monitors were used in 1978. That sentence is an inference tested in the paper, not an error attributed to another scholar. The museum’s fuller account supplies the necessary distinction. Playback order, installation plans and spectators’ attention remain separate questions requiring further evidence.
Main paper, PDF p. 8 · Supplement 2, PDF p. 7, section 4.
The seven-entry acquisition comparison was selected retrospectively. Its findings concern those records; they do not estimate institutional error rates or demonstrate that RAI outperforms competent historical practice. PDF page references above count the cover as page 1.
About this methodology record
Author: Dorian Vale · ORCID 0009-0004-7737-5094. Publisher: Museum of One. Public overview: V.1, first published 9 November 2026. The three DOI-linked publications are separate documents, each version 1.0, issued 9 November 2026.
This website is the main public introduction to RAI. The OSF companion wiki provides a related repository account, and the OSF files area holds deposited copies. Cite the specific paper or supplement when using its argument or evidence; use the overview citation below when referring to this explanation of the method.
Applications in Korean experimental art
When Documentation Becomes the Work examines the place of photographs in performance, reperformance and later collection histories through Rim Dong Sik, Kim Kulim and Sung Neung Kyung. Cite version 2.0.
Titles, Dates, and Reconstructed Works examines naming, dating and artwork identity through four cases from Korean museum and archival records. Cite version 1.0.
Cite this methodology overview
Vale, Dorian. Relational Archival Investigation (RAI). Museum of One. Methodology overview. V.1 (first public edition). https://www.museumofone.art/relational-archival-investigation/
The definition follows the author’s supplied manuscript. Its earlier draft revisions are documented in the development history below. It is not the foundational paper, a separate empirical study, or an assertion of formal peer review.
Development history and unpublished draft revisions
Development and version history
The formulation has a history.
These are unpublished manuscript revisions leading to the first public edition, V.1. They are not evidence of public deposit, peer review or a completed comparative evaluation. No manuscript is distributed through this history.
- Draft 1.2
Final documented prepublication draft
Consolidates the four commitments, strengthens the evidential demonstrations and populated research records, and develops the methodological genealogy and prospective evaluation design.
- Draft 1.1
Revised manuscript formulation
Expands the comparative literature, clarifies terminology and methodological scope, develops the relationship to PIC, and specifies operational and evaluation requirements.
- Draft 1.0
First manuscript formulation
Sets out the definition, procedure, evidential distinctions and conditions for evaluation.