The reactor anomaly and the spectrum bump ask different questions

A revised denominator changed the historical comparison

The reactor antineutrino anomaly took shape in 2011 after revised emission calculations raised predicted yields. Earlier short-distance measurements then appeared lower relative to the new reference. The observed counts had not suddenly fallen; the denominator in the comparison had changed.

Mention and colleagues’ historical reanalysis reported the observed-to-predicted ratio below. It belongs to that selected set of experiments and those calculations. It should not be carried forward as a timeless global deficit, especially when later flux models or experimental corrections change the comparison.

The ratio also requires more than an emission prediction. Exposure, target protons, interaction cross section, efficiency and backgrounds connect the predicted source to recorded counts. An absolute discrepancy can arise anywhere along that chain. A surprising ratio motivates checks; it does not identify the broken link by itself.

Sources: [1] · [2] · [3]

The bump is a local shape disagreement

Daya Bay, RENO and Double Chooz found an excess relative to adopted predictions in a restricted part of their prompt-energy spectra. The familiar “five MeV bump” names the approximate prompt-energy region. It must not be read as an exact incoming-antineutrino energy or a new monochromatic particle line.

There is no contradiction between a local excess and an integrated deficit. A spectrum can sit above its reference in one interval and below it elsewhere. Normalizing two spectra to equal areas can also hide a rate difference while leaving a shape difference visible, so the plotting convention matters.

RENO reported that the excess tracked reactor power, and several instruments saw related structure. Those checks support a reactor-related origin over a background that is constant while the cores change. They do not, by themselves, isolate a particular nuclear branch or prove that every response uncertainty is absent.

Sources: [4] · [5] · [6]

Fuel evolution tests whether one scale factor is enough

In 2017, Daya Bay followed the yield as the plutonium fission fraction evolved. A common fractional reduction of every isotope would predict a different relationship from an error concentrated in one isotope’s predicted yield. The operating reactor thus provided a way to test more than a single average.

That analysis found that the isotope contributions did not follow an equal fractional deficit relative to its reference. This created tension with a sterile-only explanation of the full rate anomaly under the tested assumptions. It did not demonstrate that no sterile state could exist in any parameter range.

Compact highly enriched research reactors provide another useful source mixture, dominated by uranium-235 fission. STEREO’s measured spectrum offers a direct isotope-specific reference. Comparing source mixtures helps investigate nuclear predictions; it is distinct from comparing distances to search for an oscillation phase.

Sources: [7] · [8] · [9]

A source-model revision and a new state predict different tests

Nuclear calculations depend on fragment yields, decay branches and corrections for transitions beyond the simplest allowed approximation. Hayes and colleagues showed that uncertain forbidden-transition treatments can matter at the scale of the historical discrepancy. Improving the error model changes how strongly an apparent mismatch can be interpreted.

A simple additional sterile state instead introduces another propagation scale. Its clearest reactor test is a spectrum that changes with distance in the predicted way, after geometry and response are included. A feature shared unchanged by nearby baselines is more naturally a source-spectrum question than that oscillation signature.

Later flux-model studies, including Giunti and colleagues’ 2021 comparison, found that the strength of the rate anomaly depends on the adopted prediction. This is why “the anomaly has disappeared” also needs qualification: agreement under one model is a result about that model and dataset, not a closure of all neutrino anomalies.

Sources: [10] · [11] · [8] · [9]

Read an anomaly as a well-specified mismatch

A useful report names the dataset, prediction, energy definition and uncertainty covariance. It then says whether the comparison tests normalization, shape, fuel dependence or distance dependence. Without those labels, two papers can appear to disagree while asking different questions of overlapping records.

Oscillation fits based on near–far comparisons can remain robust when a shared spectral prediction is imperfect. Absolute flux measurements are more exposed to that imperfection because their purpose is to test it. The same experiment can therefore deliver strong mixing evidence and an unresolved source-spectrum investigation.

The next article follows the distance test directly. That is the productive role of an anomaly: it suggests alternatives that make different predictions and invites new controls. Counting the word as evidence for a new particle skips the part of the science that can distinguish those alternatives.

Sources: [12] · [7] · [8] · [13]

Try it in the Lab

Primary sources & revision

1. Mention et al. · The Reactor Antineutrino Anomaly (2011)
2. Mueller et al. · Improved Predictions of Reactor Antineutrino Spectra (2011)
3. Patrick Huber · On the determination of anti-neutrino spectra from nuclear reactors (2011)
4. Daya Bay Collaboration · Measurement of the Reactor Antineutrino Flux and Spectrum (2015/2016)
5. Seon-Hee Seo for RENO · New Results from RENO and The 5 MeV Excess (2014)
6. Double Chooz Collaboration · Improved measurements of the neutrino mixing angle (2014)
7. Daya Bay Collaboration · Evolution of the Reactor Antineutrino Flux and Spectrum (2017)
8. STEREO Collaboration · Uranium-235 spectrum and sterile-neutrino test (Nature, 2023; arXiv v3, 2024)
9. PROSPECT Collaboration · Final PROSPECT-I Short-Baseline Oscillation Search (2024 submission; 2025 publication)
10. Hayes et al. · Systematic Uncertainties in the Analysis of the Reactor Neutrino Anomaly (2013/2014)
11. Giunti, Li, Ternes & Xin · Reactor antineutrino anomaly in light of recent flux model refinements (2021/2022)
12. Daya Bay Collaboration · Observation of electron-antineutrino disappearance (2012)
13. Particle Data Group · Statistics (2025)

First published 2026-10-10; last revised 2026-10-10. Original explanatory text and diagrams by Sandbox Physics. Illustrations are schematic; no experimental event records are reproduced here. This is an independent educational publication, not an official Nobel or experiment collaboration publication.

Continue the story