Corrections & Amendments
A finding does not stop being subject to scrutiny the moment it is published. When a later test, a Devil’s Advocate pass, or a registry change forces us to revise something we have already put on the record, we amend the affected page in place — and we log the amendment here.
This page is the editorial counterpart to the Killed Signals list. Killed Signals tracks hypotheses retired at the research level; this page tracks corrections made to material already published. A programme that quietly edits its own history is not doing science. Every entry below links to the page it changed.
| Date | What changed | Affected page |
|---|---|---|
| 2026-04-19 | clark_english_prices moved to SUSPENDED (validator never run); the cited 4.4-year spectral peak is now a single-test curiosity, not a confirmed finding. Affects the Fimbulwinter and Ancient Precision papers. | English wheat-price periodicity bulletin · Fimbulwinter Hypothesis · Ancient Precision Archive |
| 2026-04-19 | Wealth-concentration bulletin: confidence downgraded after a post-publish Devil’s Advocate review found the trend not separable from equity-market appreciation. | US wealth-concentration bulletin |
| 2026-04-21 | “Why 2029?” page retracted — the four-chain convergence-score framework was retired after multiple underlying signals moved to NOISE. Page retained as a historical record. | Why 2029? |
| 2026-04-24 | AMOC bulletin: confidence amendment following review of trend-fit and deseasonalisation method. | AMOC transport-decline assessment |
| 2026-04-25 | honey_antibacterial citation corrected: Robson 2009 replaced with Gethin & Cowman 2008 (PMID 18666717). | Honey as an antimicrobial agent bulletin |
| 2026-04-25 | Fimbulwinter paper: effect-size methodology amendment. | Fimbulwinter Hypothesis |
| 2026-05-17 | Week One Retrospective: headline confidence reframed from 0.95 to Moderate. | Observatory Week One Retrospective |
| 2026-05-17 | Ancient Precision Archive: population-level reconciliation — the broad “ancient knowledge as a privileged source” framing reconciled against the killed-at-population-level battery result. | The Ancient Precision Archive |
| 2026-08-03 | nilometer_enso_1300yr moved to CONTESTED: its earlier CONFIRMED 4/4 was a fabricated (seeded) battery tuned to the published modern correlation; re-run on real data the medieval ENSO-Nile correlation is a robust null. The nilometer leg is withdrawn from the Cosmic-Ray causal-chain paper and flagged as contested in the Week One Retrospective. An earlier in-place amendment that had certified nilometer as CONFIRMED is superseded. | Cosmic-Ray Consilience · Observatory Week One Retrospective |
| 2026-08-03 | fimbulwinter_volcanic_clustering moved to CONTESTED: rewritten on the eVolv2k volcanic-sulfate record, large eruptions are approximately Poisson-distributed, not clustered – the paper’s central doublet/clustering thesis is refuted. The volcanic-cooling mechanism and the 536/540 CE anchor survive. The prior “clustering claim unaffected” amendments are superseded; paper retained as a historical record. | The Fimbulwinter Hypothesis |
Where an amendment changes a published bulletin or paper, the affected page itself carries an in-place amendment note; this list is the network-wide index of those changes.