A correction amends specific elements of a published work — metadata, text, figures, or acknowledgements — while affirming that the core findings remain reliable.

An erratum notifies readers of errors introduced in production or publication that affect accuracy but may not invalidate conclusions. A corrigendum notifies readers of errors attributable to authors that require correction in the published record.

An expression of concern is an interim notice that credible questions about integrity or reliability are under investigation. It alerts readers without concluding guilt or innocence.

A withdrawal removes a work from active scholarly circulation before or after publication, with visibility depending on stage and cause. A retraction formally invalidates or withdraws published findings when reliability or ethics requires it, with a permanent linked notice.

Version history records each published state, linked correction, and retraction notice so readers can trace how the record evolved. Persistent identifiers resolve to current metadata with access to prior versions where archived.

Transparency requires public, intelligible notices that explain what changed and why. Fairness requires that authors be informed and heard before consequential actions where practicable. Due process requires documented investigation, evidence review, and proportionate outcomes.

Scientific integrity demands that unreliable or unethical work not continue to circulate as valid science. The permanent scholarly record is preserved — retractions and corrections are visible rather than silent deletions except where law mandates restriction.

Accountability applies to authors, editors, and the Platform. Decisions are attributed in audit logs; appeals address procedural error and new evidence.

Examples include typographical errors, incorrect author names or affiliations, missing or erroneous funding acknowledgements, metadata corrections such as titles or keywords, minor factual clarifications that do not alter conclusions, and replacement of figures when the scientific content remains valid but display was flawed.

Corrections should describe the error, the change made, and the date of correction. The original publication remains accessible with a linked correction notice. Authors may initiate corrections; editors may require them when errors are identified during review of reader reports or internal checks.

Corrections are not appropriate when errors are so fundamental that the work's conclusions cannot stand — such cases may require errata with broader scope, corrigenda, or retraction.

Errata typically address publisher-side or production errors — formatting, typesetting, or metadata introduced after author approval. Corrigenda address author-side errors in content, data presentation, or analysis that readers must know to interpret the work correctly.

Both notice types link to the original article, state what was wrong, what is correct, and whether conclusions are affected. The original version remains in the archive for citation archaeology; the version of record reflects corrected material where applicable.

When errors affect only supplementary materials, notices may target those files while the main article receives a cross-link.

Expressions of concern are appropriate when institutions, funders, or credible third parties report possible fabrication, plagiarism, or ethics violations; when internal editorial review identifies serious unresolved questions; or when legal or regulatory investigations are pending.

Notices should be factual, avoid prejudging outcomes, and state that investigation is in progress. They are temporary markers replaced by correction, retraction, or removal of concern when resolved.

Authors should be notified before publication of an expression of concern except where immediate reader protection or legal obligation requires expedited notice.

Pre-publication withdrawal occurs when authors withdraw submissions before final publication — unpublished manuscripts are not released publicly. Post-publication withdrawal may occur when authors discover fatal flaws before formal retraction processes complete, or when mutually agreed removal is appropriate for works published in error.

Administrative withdrawal may occur when duplicate records, test publications, or policy violations require removal without full retraction proceedings. Author-requested withdrawal is honored when no integrity investigation opposes it and legal obligations permit.

Withdrawn post-publication works retain a public notice explaining withdrawal where scholarly record principles apply, rather than vanishing without trace.

Grounds include major honest errors that invalidate conclusions when correction is insufficient; fabrication or falsification of data; plagiarism or duplicate publication; image manipulation that misrepresents findings; authorship fraud; undisclosed conflicts or ethics violations that undermine validity; fake peer review; and legal requirements prohibiting continued distribution.

Retractions are intended to preserve the scholarly record, not erase it. Retracted articles remain accessible with a prominent retraction notice, retraction metadata, and updated trust labels. Citations should point to the retracted status.

Only authorized editors and administrators may retract publications. Retraction statements name the reason, parties notified, and whether findings are wholly or partially invalid.

Authors should receive notice of proposed corrections that materially affect conclusions, expressions of concern, and retractions with opportunity to respond before publication of notices when time and safety permit.

Authors may provide explanatory statements, corrective data, or institutional investigation outcomes. Good-faith cooperation may result in correction rather than retraction when reliability can be restored.

Appeals of retraction and major integrity findings may be submitted under Article XIII. Author responses may be linked to notices where appropriate and lawful.

Editors gather evidence from authors, complainants, institutions, and experts without predetermined outcomes. Conflicts of interest require recusal or managing editor reassignment.

Confidentiality protects whistleblowers, ongoing investigations, and unpublished allegations until notices are issued. Written rationale supports appeals and institutional review.

Consistency means similar cases receive similar treatment across publication types, without favoritism based on author prominence or institution.

Evidence may include original data, analysis files, institutional investigation reports, third-party image analysis, and correspondence. Expert consultation supports technical disciplines where editors lack specialized expertise.

Institutional consultation may occur when employers or funders investigate parallel proceedings; Inquisra may await outcomes or proceed independently when public health or reader protection requires.

Decisions are made by authorized editors or integrity administrators. Documentation includes evidence summary, parties heard, policy applied, and notice text.

Version timelines show submission revisions, published versions, corrections, and retraction events. Readers can identify the current version of record and access prior states where archived.

Correction and retraction notices carry their own metadata and link bidirectionally to affected versions. Identifiers resolve to current status while metadata registers relationships to prior versions.

Silent overwrite of published content is prohibited. Integrity actions always leave an auditable trail.

Retraction banners, correction labels, and expression-of-concern flags appear on publication pages and in discovery metadata where technically feasible. Notice text explains causes in language accessible to non-specialists when possible.

The scholarly record includes failures as well as successes — hiding retractions defeats the purpose of correction. Restricted access applies only where law, privacy, or safety requires.

Trust metadata reflects current integrity status so automated systems and readers can detect retracted work.

Appeals must state specific grounds — procedural error, new exonerating evidence, disproportionate sanction, or factual mistake in the notice. Appeals are not venues to relitigate scientific disagreement with editorial rejection.

Independent review may be assigned to managing editors or integrity administrators not involved in the original decision. Outcomes may uphold, modify, or reverse actions with updated public notices.

Appeal deadlines and formats are communicated in decision correspondence. Frivolous or harassing appeals may be declined with explanation.

Undisclosed conflicts may warrant correction or retraction when they invalidate interpretation. AI misuse under the Responsible AI Charter may trigger integrity review. Editorial Ethics governs investigation authority and enforcement ladders.

Publication type requirements and reporting guidelines inform whether errors constitute minor correction or fundamental unreliability. Copyright and licensing charter provisions govern retention of retracted records.

Integrated governance ensures consistent treatment of related issues — plagiarism, duplicate publication, and image manipulation connect to this charter's correction and retraction tools.

The Platform provides infrastructure for versioning, notices, audit logs, and public trust labels — not merely policy text. Integrity is operationalized in product design and editorial workflow.

Inquisra will not remove retractions under commercial pressure, will cooperate with institutions within legal bounds, and will update this charter as community standards evolve.

Researchers, editors, and readers share responsibility for using integrity tools honestly: corrections for errors, retractions for unreliability, and appeals for procedural justice.