Search for eV Sterile Neutrinos -- The STEREO Experiment [Blois 2019]

Stefan Schoppmann

REVIEW 3 major objections 3 minor 9 references

Search for eV Sterile Neutrinos -- The STEREO Experiment [Blois 2019]

T0 review · 3 major / 3 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Using 43,400 inverse-beta-decay events from six cells at 9-11 m, STEREO finds no eV-scale sterile neutrino oscillation and rejects the Reactor Antineutrino Anomaly's best-fit point at 99% C.L.

desk verdict A real new STEREO phase-II result with a clean flux-independent shape analysis, but the proceedings leaves a day-count inconsistency and the background-extrapolation closure unshown. read the letter →

arxiv 1909.01017 v1 pith:RTKRZBDD submitted 2019-09-03 hep-ex physics.ins-det

Stefan Schoppmann This is my paper · ORCID

classification hep-exphysics.ins-det

keywords sterileneutrinosearchreactorantineutrinoanomalyinversebetadecaysegmenteddetectorflux-model-independentanalysisoscillationexclusioneV-scalegadolinium-loadedscintillator

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

STEREO reports phase-II data from a detector segmented into six identical cells placed 9.4-11.1 m from a compact research reactor core, with 119 days of reactor-on and 211 days of reactor-off exposure. Instead of comparing measured antineutrino rates to an absolute flux prediction, the analysis compares the six cells against a common per-energy-bin normalization, so the oscillation search is independent of reactor flux uncertainties. The data are compatible with the no-oscillation hypothesis, with a p-value of 0.4, and the best-fit point of the Reactor Antineutrino Anomaly is rejected at 99% C.L. The result matters because it weakens the case that the short-baseline reactor anomaly is caused by an eV-scale sterile neutrino, pointing instead toward errors in the predicted reactor flux.

What carries the argument

The load-bearing object is the six-cell target volume: six optically separated cells measure the same neutrino source at six baselines, so an oscillation signal appears as a relative distortion across cells rather than as a change in absolute rate. The analysis removes all absolute flux information by introducing a free normalization $\phi_i$ for each energy bin, which rescales the expected spectrum across all cells to match the measured average. The remaining cell-to-cell differences are fitted for oscillation parameters, with nuisance parameters covering energy-scale and per-cell normalization uncertainties. This construction makes the search independent of any reactor flux prediction and insensitive to correlated systematic uncertainties.

What would settle it

Split the reactor-off data into periods with the largest spread in temperature and atmospheric pressure, apply the same corrections, and check whether the corrected spectra agree; if they differ by more than the assigned systematic uncertainty, the background extrapolation is biased. A reactor-on/off cycle during a strong weather front would provide the same test on real data.

Watch

Extended reading notes

Core claim

The central result is the phase-II oscillation measurement: from 43,400 inverse-$\beta$-decay candidates distributed over six cells, the relative rates and energy spectra across baselines 9.4-11.1 m show no distortion of the kind an eV-scale sterile neutrino would produce. A binned $\chi^2$ fit with free flux normalizations per energy bin yields a p-value of 0.4 for the null hypothesis, and the best-fit point of the Reactor Antineutrino Anomaly in the $\sin^2(2\theta_{ee})$--$\Delta m^2_{41}$ plane is excluded at 99% C.L. An independent covariance-matrix version of the fit confirms the exclusion region.

Load-bearing premise

The load-bearing premise is that reactor-off background spectra, after corrections for temperature and atmospheric pressure, accurately represent the time-varying background during reactor-on data taking; the proceedings states this correction is needed but does not show a quantitative closure test, so a bias here could distort the six-cell spectra and create or hide an oscillation signature.

Editorial extensions

If this is right

If the result is correct, the specific parameter region preferred by the Reactor Antineutrino Anomaly is excluded as an explanation of the STEREO data.
Short-baseline reactor searches can be conducted without knowing the absolute reactor flux, as long as the detector is segmented and the relative response is calibrated.
The 6.5% reactor flux deficit is more plausibly a problem with flux prediction than a sterile-neutrino oscillation at baselines near 10 m.
Combining phase-I and phase-II data will extend the exclusion region and sharpen comparisons with other short-baseline experiments.
The null result adds tension with shape-only global combinations that currently prefer a sterile neutrino, so final systematics from other experiments will be decisive.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The free-normalization construction could be reused by any segmented detector whose cell baselines are known precisely, even when the absolute flux is unknown.
If the reactor-off background correction passes a dedicated closure test, the sensitivity of the method should continue to grow roughly as the square root of the number of inverse-beta-decay candidates, making the combined phase-I+II dataset a natural next test.
A spectral fit that also uses the absolute rate information, once flux predictions improve, could distinguish between a flat flux deficit and an energy-dependent distortion.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is a proceedings write-up of STEREO's phase-II sterile-neutrino search, and it contains a genuinely new result that appears sound. Using only relative comparison of six identical cells, with per-energy-bin normalizations that absorb the absolute flux prediction, they find the null hypothesis perfectly compatible (p=0.4) and reject the reactor antineutrino anomaly best-fit point at 99% C.L. That's worth knowing about, and it is not the usual claim-versus-prediction torture.

What's actually new: the phase-II dataset (199 days reactor-on, 211 days off, 43,400 IBD candidates) and the updated exclusion contour. The method is not new—it's an extension of the phase-I shape analysis—but this is the first time the phase-II data alone are presented. The analysis is coherent: the χ² with free φ_i per energy bin really does make the test independent of the absolute flux, and they back the raster-scan result with an independent covariance-matrix treatment plus a pseudo-experiment p-value. That's good practice.

Soft spots, in order of concern. One: there's a concrete data-statement inconsistency—the abstract and Figure 2 legend say 119 days of reactor-on, while Section 3 says phase-II spans 199 days. This needs to be fixed; it's likely a leftover from phase-I (which used 119 days), but it's sloppy in a paper whose central claim depends on the dataset. Two: the background model is the least documented load-bearing input. The reactor-off spectra are corrected for temperature and atmospheric pressure, but the proceedings gives no magnitude, no residual plots, and no closure test. Since the analysis relies on cell-to-cell differences at fixed energy, any residual that varies across cells or energy could bias the oscillation test. I don't see evidence of a problem—the p=0.4 null consistency is reassuring—but a full paper needs to show that closure. Three: there's no systematic-error table in this proceedings, so the 99% C.L. rejection can't be fully audited from this document alone.

The circularity concern is a non-issue: the RAA parameters are used only as a target to exclude, not as an input, and the free φ_i make the test flux-independent.

Bottom line: this is a solid experimental result that deserves a serious referee, but as a proceedings it needs a corrected day count and at least a summary of the background-correction closure and systematics before I'd treat it as the definitive phase-II statement. If the authors fix the typo and back the background treatment, it's a clean contribution to the short-baseline puzzle.

Referee Report

3 major / 3 minor

Summary. This proceedings paper reports the STEREO experiment's phase-II search for eV-scale sterile neutrinos using the compact ILL reactor. The detector consists of six identical liquid-scintillator cells at baselines between 9.4 and 11.1 m. The analysis is deliberately independent of any absolute flux prediction: per-energy-bin normalizations φ_i absorb all absolute rate information, so only relative cell-to-cell distortions are used. With 13 energy bins and six cells, the authors perform a χ² fit with per-cell uncorrelated energy-scale and normalization nuisance parameters plus a common energy-scale parameter. They report compatibility with the null oscillation hypothesis (p = 0.4 from pseudo-experiments) and claim to reject the best-fit point of the Reactor Antineutrino Anomaly (RAA) at 99% C.L. They also state that an independent covariance-matrix method confirms the raster-scan results.

Significance. If the analysis is correct, this is an important result: it excludes the RAA's preferred sterile-neutrino parameter point with a flux-model-independent method, a significant advance over flux-dependent analyses. The paper's design is strong: the φ_i free parameters remove reliance on reactor flux predictions, pseudo-experiments provide a non-Gaussian null test, and an independent covariance-matrix cross-check guards against mis-modeling in the χ² form. The result is also timely, as other short-baseline experiments (DANSS, NEOS, NEUTRINO-4) currently disagree on the sterile-neutrino interpretation. However, the strength of the exclusion claim depends critically on the correctness of the reactor-off background extrapolation and on a clear statement of the data sample, both of which are incompletely documented in this proceedings.

major comments (3)

[Section 3, background correction paragraph] The analysis relies on reactor-off background spectra corrected for temperature and atmospheric pressure, but no closure test, residual distribution, or magnitude of the correction is reported. The text states only that 'it is not possible to apply the background spectra directly as they scale, e.g. with temperature or atmospheric pressure. Thus, spectra are corrected for those effects exploiting different temperature and atmospheric pressure settings during reactor-off phases.' Because the per-energy-bin φ_i parameters absorb all absolute rate information, any cell-dependent or energy-dependent residual in the background extrapolation enters directly as a fake oscillation signature. A residual of order 1% per cell-energy bin is comparable to the expected relative oscillation signal and could shift the 99% C.L. exclusion. Please provide quantitative validation of the background correction, for example closure tests on reactor-off data split by environmental conditions, residual plots as a function of cell and energy, or a specific reference to a detailed analysis paper where such tests are shown.
[Abstract, Section 3, and Figure 2 legend] The dataset is described inconsistently: the abstract says '119 days of reactor turned on and 211 days of reactor turned off', Section 3 says 'Phase-II spans 199 days of reactor-on and 211 days of reactor-off', and the Figure 2 legend says 'Exclusion (119 days)'. The exact exposure and the relationship between '119 days' and '199 days' must be reconciled, since the statistical power of the exclusion and the quoted p-value directly depend on the integrated live time. Please correct the typo and state the final exposure unambiguously.
[Section 3 and Figure 2] The abstract and text claim that the best-fit point of the RAA is 'rejected at 99% C.L.', but Figure 2 only displays a 90% C.L. exclusion contour. The reader cannot directly verify the central claim from the figure. Please either show the 99% C.L. contour, or provide the numerical confidence level or p-value (e.g., from the scan at the RAA best-fit Δm² value) so that the 99% rejection statement is supported by the displayed data.

minor comments (3)

[References] Reference 3 has a typo: 'arXiv1906.01739' should read 'arXiv:1906.01739'.
[Section 3, acceptance statement] The phrase 'The overall acceptance in the TG is 61% showing a slight correlation with energy of 4% between 3 and 8 MeV' would benefit from a definition of whether the 4% is a relative or absolute variation, and from a reference to the detailed efficiency study.
[Table 1] The table of selection cuts mixes formatting conventions (e.g., 'µs' and 'mm'); this is a minor presentational issue but should be harmonized in the final version.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: flux-shape-free relative cell-rate test is self-contained.

full rationale

I find no circular step in this proceedings. The central claim is a relative-rate comparison among the six detector cells. In Eq. (1), the per-energy-bin free normalizations φ_i are common to all cells, so they absorb the absolute reactor flux prediction while leaving any cell-dependent spectral distortion, such as that from oscillations, to be tested by the data. The null hypothesis is assessed by comparing the resulting Δχ² to pseudo-experiments, and the best-fit point of the Reactor Antineutrino Anomaly is used only as a fixed external target for exclusion, not fitted from the STEREO data. The temperature and atmospheric-pressure corrections to reactor-off backgrounds represent a systematic assumption, but they are not a case of the analysis predicting its own input; at most they are a possible source of bias, not circularity. Self-citations to earlier STEREO publications describe detector construction, calibration, and simulation, and they are not the load-bearing argument for the null-oscillation result. Therefore, no circularity is apparent and the score is 0.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The analysis relies on standard statistical assumptions, on reactor-off backgrounds being transportable to reactor-on conditions, and on cell-to-cell response being modeled by nuisance parameters. No new particles, forces, or entities are introduced.

free parameters (4)

phi_i (per-energy-bin common normalization) = not reported (analysis free parameter)
Absorbs the absolute flux prediction for each of the 13 energy bins, making the analysis shape-only and removing dependence on reactor flux models.
alpha_EscaleU_l (per-cell uncorrelated energy scale) = not reported (analysis free parameter)
Uncorrelated per-cell energy-scale nuisance parameter in Eq. (1), allowed to account for relative calibration differences between cells.
alpha_EscaleC (common energy scale) = not reported (analysis free parameter)
Common energy-scale nuisance parameter in Eq. (1), representing correlated energy-scale uncertainty across the detector.
alpha_NormU_l (per-cell uncorrelated normalization) = not reported (analysis free parameter)
Uncorrelated per-cell normalization nuisance parameter in Eq. (1); together with phi_i it removes absolute rate information from the fit.

assumptions (3)

domain assumption Reactor-off IBD background spectra, after temperature and pressure corrections, represent the background during reactor-on data taking.
Section 3 estimates the background distribution from reactor-off phases and corrects for temperature/pressure; if incorrect, the shape comparison is biased.
domain assumption The six target cells have identical response up to per-cell nuisance parameters, and the source geometry is known well enough that relative cell fluxes depend only on L/E oscillation.
Section 2 and Eq. (1) assume M_l,i can be modeled with common oscillation parameters plus per-cell nuisance terms.
standard math The pseudo-experiment distribution and the fixed Delta m^2 raster-scan chi-square procedure are valid frequentist methods for the null and exclusion tests.
Section 3 uses pseudo-experiments to test no-oscillation and a raster scan over fixed mass splittings to define the exclusion region.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Search for eV Sterile Neutrinos -- The STEREO Experiment [Blois 2019]." pith.science (2026). https://pith.science/paper/RTKRZBDD

@misc{pith2026190901017,
  author       = {Pith},
  title        = {Pith review of: Search for eV Sterile Neutrinos -- The STEREO Experiment [Blois 2019]},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RTKRZBDD}},
  note         = {Machine review of arXiv:1909.01017}
}

read the original abstract

The STEREO experiment is designed to test the hypothesis of light sterile neutrinos being the cause of the Reactor Antineutrino Anomaly. It measures the antineutrino energy spectrum from the compact core of the ILL research reactor in six identical detector cells covering baselines between 9 and 11 m. Results from 119 days of reactor turned on and 211 days of reactor turned off are reported. Using a direct comparison between neutrino interaction rates of all cells, independent of any flux prediction, we find compatibility with the null oscillation hypothesis. The best fit point of the Reactor Antineutrino Anomaly is rejected at 99% C.L.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

9 extracted references · 7 canonical work pages

[1]

Mention et al., Phys

G. Mention et al., Phys. Rev. D 83, 073006 (2011)

2011
[2]

Almaz´ an et al

H. Almaz´ an et al. (Stereo Collaboration), Phys. Rev. Lett. 121, 161801 (2018)

work page 2018
[3]

B¨ oser et al., arXiv1906.01739, [hep-ex] (2019)

S. B¨ oser et al., arXiv1906.01739, [hep-ex] (2019)

arXiv 2019
[4]

Allemandou et al., JINST 13, P07009 (2018)

N. Allemandou et al., JINST 13, P07009 (2018)

work page 2018
[5]

C. Buck, B. Gramlich, M. Lindner, C. Roca and S. Schoppmann, JINST 14, P01027 (2019)

work page 2019
[6]

Litaize, O

O. Litaize, O. Serot and L. Berge, EPJ A 51, 177 (2015),

work page 2015
[7]

Improved STEREO simulation with a new gamma ray spectrum of excited gadolinium isotopes using FIFRELIN

H. Almaz´ an et al., arXiv1905.11967, [physics.ins-det] (2019)

work page Pith review arXiv 2019
[8]

Almaz´ an et al., Zenodo2653786, [Data set] (2019)

H. Almaz´ an et al., Zenodo2653786, [Data set] (2019)

work page 2019

Show all 9 references

[9]

Lhuillier et al

D. Lhuillier et al. ( Stereo Collaboration), Institut Laue-Langevin (ILL) DOIs:, 10.5291/ILL-DATA.ST-6, 10.5291/ILL-DATA.ST-7, 10.5291/ILL-DATA.ST-8, 10.5291/ILL-DATA.ST-9, 10.5291/ILL-DATA.ST-10, 10.5291/ILL-DATA.ST-11, 10.5291/ILL-DATA.ST-12 (2016–2019)

doi

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Mention et al., Phys

G. Mention et al., Phys. Rev. D 83, 073006 (2011)

2011

[2] [2]

Almaz´ an et al

H. Almaz´ an et al. (Stereo Collaboration), Phys. Rev. Lett. 121, 161801 (2018)

work page 2018

[3] [3]

B¨ oser et al., arXiv1906.01739, [hep-ex] (2019)

S. B¨ oser et al., arXiv1906.01739, [hep-ex] (2019)

arXiv 2019

[4] [4]

Allemandou et al., JINST 13, P07009 (2018)

N. Allemandou et al., JINST 13, P07009 (2018)

work page 2018

[5] [5]

C. Buck, B. Gramlich, M. Lindner, C. Roca and S. Schoppmann, JINST 14, P01027 (2019)

work page 2019

[6] [6]

Litaize, O

O. Litaize, O. Serot and L. Berge, EPJ A 51, 177 (2015),

work page 2015

[7] [7]

Improved STEREO simulation with a new gamma ray spectrum of excited gadolinium isotopes using FIFRELIN

H. Almaz´ an et al., arXiv1905.11967, [physics.ins-det] (2019)

work page Pith review arXiv 2019

[8] [8]

Almaz´ an et al., Zenodo2653786, [Data set] (2019)

H. Almaz´ an et al., Zenodo2653786, [Data set] (2019)

work page 2019