One-sentence brief
A number can be internally correct and still misleading when its unit or inclusion universe is hidden.
NUCLEAR TESTING HISTORY AND EVIDENCE
A unit guide for comparing official test totals, device counts, event rows, grouped detonations, peaceful explosions, safety experiments, and wartime nuclear detonations.
ORIENTATION
A number can be internally correct and still misleading when its unit or inclusion universe is hidden.
WORKING BRIEF
An event-level catalog row is a source convention. It may aggregate multiple devices or represent a special experiment rather than a standard weapons test.
A device count answers how many devices detonated, not how many test dates, operations, or independently verified events occurred.
Combat use is legally, historically, and ethically distinct from a test program and must remain a separate event class.
Compare like with like, or publish a mapping table that explains the conversion and unresolved differences.
COMPLETE DOSSIER
Terms are defined for this site’s evidence method, not as universal legal or clinical definitions.
| Claim state | Acceptable wording | Prohibited implication |
|---|---|---|
| Exact dataset count | “The exact supplied dataset contains…” with unit and revision. | Historical completeness, independent verification, or identical scope with another total. |
| Source-authored observation or estimate | “The source reports/estimates…” with population, period, and uncertainty. | Universal fact, individual attribution, or causation beyond the study design. |
| Modeled relationship | MODELED RELATIONSHIP — NOT A FORECAST Explain inputs, assumptions, outputs, validation, and transfer. | Live forecast, guaranteed arrival, or automatic action. |
| Projection or proposal | SOURCE-AUTHORED PROJECTION — NOT A SITE FORECAST / POLICY PROPOSAL — NOT IMPLEMENTED HERE | Site forecast, deployment claim, authority, or inevitability. |
Are counting unit, population, period, geography, and comparison explicit?
Pass condition: The claim can be reproduced without silently changing its universe.
Are source instances separated from unique evidence origins?
Pass condition: Duplicate and derivative files are not counted as corroboration.
Is observation, estimate, model, projection, or proposal status visible with a review date?
Pass condition: Mutable claims have currentness and expiration triggers.
Are correction dependencies and authority limits explicit?
Pass condition: No result implies approval or action and every material change has a propagation path.
RESEARCH EDITION
This page follows the public method for provenance, confidence, source independence, alternative accounts, limitations, review state, and visible correction.
CONTINUE