Papex

Open problem · Physics · Open

Neutrino mass hierarchy

Are the three neutrino masses arranged in the normal or the inverted order?

Summary

Oscillation experiments measure only the differences between the squared masses of the three neutrinos, which leaves two possible orderings, normal and inverted; which one nature follows is not yet established.

Why it is open

Each of the three neutrino flavours (electron, muon and tau) is a mixture of three states with definite masses, ν1, ν2 and ν3 [1]. Solar neutrinos, whose oscillations are shaped by the dense matter of the Sun, show that ν2 is heavier than ν1, where ν1 is defined as the state with the largest electron-neutrino component [1]. Whether ν3 is heavier than both (the normal hierarchy, or ordering) or lighter (the inverted one) has not been determined [1, 5]. In vacuum, oscillations between two flavours depend on the size of a squared-mass difference but not on its sign [1]. Xin Qian and Petr Vogel describe two ways round this: the matter effect on neutrinos that travel long distances through the Earth, and the small difference between the squared-mass splittings Δm²31 and Δm²32, seen in the disappearance of reactor antineutrinos; both effects are subtle [1].

Knowing the ordering would "strengthen, or eliminate, roughly half of the proposed models" of neutrino mass, Qian and Vogel write [1]. It sets a floor on the sum of the three masses, which cosmology measures: 0.059 eV for the normal and 0.10 eV for the inverted ordering [1, 9]. And if the ordering is inverted, the effective mass that sets the rate of neutrinoless double-beta decay, the best test of whether neutrinos are Majorana particles (their own antiparticles), is at least 14 meV with the oscillation data of 2015, while the normal ordering allows zero [1].

No body sets a bar for a resolution. Planned experiments state their aims as significances: DUNE projects resolving the ordering at 5σ (five standard deviations) for all values of the CP-violating phase δCP [3], and Qian and Vogel consider it desirable that several methods be used, so that the results are complementary [1]. They place the ordering within the search for a model that relates neutrino masses and mixing to the charged-lepton masses, and possibly to the quarks [1]; that wider puzzle has its own page, Generations of matter.

What is known

Atmospheric neutrinos (2024). Super-Kamiokande's complete pure-water data set of atmospheric neutrinos, from April 1996 to July 2020 (6,511.3 live days), with the mixing angle θ13 constrained by reactor experiments, "favors the normal mass ordering at the 92.3% level" [4].

Accelerator experiments (2025–2026). NOvA and T2K send neutrino beams over hundreds of kilometres and measure muon-neutrino disappearance and electron-neutrino appearance [8, 10]. After ten years, NOvA's data "show a mild preference for the normal mass ordering with a Bayes factor of 2.4", meaning that the normal ordering is "2.4 times more probable"; with a constraint based on Daya Bay reactor data the factor rises to 6.6 [10]. The first joint analysis of NOvA and T2K, published in October 2025, found "no strong preference for either mass ordering", but showed that if the inverted ordering were true, its results "would provide evidence of CP symmetry violation in the lepton sector" [8]. A joint analysis of Super-Kamiokande atmospheric and T2K beam data found "a limited preference for the normal ordering with a 1.2σ exclusion of the inverted ordering" [6].

Global fits (2024–2025). Using data available in September 2024, the NuFIT 6.0 analysis by Ivan Esteban and colleagues found "different tendencies": T2K and NOvA appearance data each favour the normal ordering but agree better with each other under the inverted ordering, while the combined disappearance data prefer the normal ordering [5]. With long-baseline, reactor and IceCube atmospheric data, both orderings fit almost equally well; only when Super-Kamiokande's atmospheric data are added does the global fit prefer the normal ordering, with Δχ² = 6.1 (a difference in goodness of fit) [5]. In 2025 Francesco Capozzi and colleagues found "a relatively weak preference for NO versus IO (at 2.2σ)" [7].

Cosmology (2025). Combining baryon acoustic oscillations (a clustering feature used to measure cosmic distances) from the Dark Energy Spectroscopic Instrument (DESI) with cosmic microwave background data, Willem Elbers and colleagues found that the sum of neutrino masses is below 0.0642 eV at 95% confidence in the standard cosmological model, ΛCDM [9]. Together with oscillation constraints this gave a Bayes factor of 10 for the normal ordering, "substantial evidence" on the Jeffreys scale, under that model [9]. They also report that their constraints are "in tension with the lower limits derived from neutrino oscillations", and that with a time-evolving dark energy the limit relaxes to 0.163 eV, "relaxing the neutrino tension" [9]. See also Dark energy.

Reactors: JUNO (2025–2026). The Jiangmen Underground Neutrino Observatory (JUNO) is a 20-kilotonne liquid-scintillator detector 52.5 km from several reactor cores, designed to resolve the interference pattern of reactor neutrinos; it was completed in August 2025 [12]. Its first 59.1 days of data improved the precision of the solar parameters sin²θ12 and Δm²21 by a factor of 1.6 over all previous measurements combined [12]. Going beyond the collaboration's published analysis, Esteban and colleagues found in 2026 that these data, combined with other determinations of the larger squared-mass splitting, give "a slight preference for Normal Ordering", with a p-value for the inverted ordering of 2–2.6% (2.2σ–2.3σ); adding the information already in the global data gives Δχ² = 4.6, or 9.4 with Super-Kamiokande and IceCube atmospheric data [11]. They call the study "exploratory", to be "corroborated by official analyses" [11].

Recent claims. On 29 September 2026 the JUNO collaboration released an analysis of 207.2 live days that measures Δm²31 to about 1% [14]. Combined with Δm²31 constraints from long-baseline accelerator experiments, it "favored normal ordering over inverted ordering, with a one-sided significance of at least 2.1σ across all possible CP-violating phases", and a Bayesian analysis gave a Bayes factor of 10.18 [14]. It is a preprint, not peer reviewed, with no journal reference listed on 16 October 2026 [14].

Main approaches

Long-baseline accelerators. Matter in the Earth enhances the appearance of electron neutrinos for the normal ordering and suppresses it for the inverted one, with the opposite effect for antineutrinos, and the effect grows with the baseline [2]. Strongest evidence: NOvA's preference for the normal ordering, strengthened by a reactor constraint [10]. Main difficulty: T2K and NOvA appearance data agree better with each other under the inverted ordering [5], and their joint fit shows no strong preference [8]. DUNE's 2020 study projects a 5σ determination for all δCP values after two years of running with the nominal detector and beam [3].

Atmospheric neutrinos. Neutrinos crossing the Earth undergo a resonant oscillation, for neutrinos if the ordering is normal and for antineutrinos if it is inverted [1]. Strongest evidence: Super-Kamiokande's 92.3% preference [4]. Main difficulty: the atmospheric flux mixes flavours and neutrinos with antineutrinos, which detectors must tell apart or account for through their different fluxes and cross sections [1]; in the joint fit with T2K, which pulls the mixing angle θ23 towards its upper octant (above 45 degrees), the atmospheric constraint on the ordering became weaker [6]. Hyper-Kamiokande's 2018 design report expects better than 3σ after five years by combining atmospheric and beam data, as matter effects over its 295 km beam are small [2].

Reactors, alone and with accelerators. Precise reactor measurements expose the small difference between Δm²31 and Δm²32 [1], and comparing reactor and accelerator values of Δm²31 tests the ordering [11, 14]. Strongest evidence: JUNO's 1% measurement and its 2.1σ combination with accelerator data [14]. Main difficulty: JUNO's first 59.1 days alone had "negligible sensitivity" to the ordering [11]; with limited statistics its fit still has several local minima, and Daya Bay data are used to pick the physical one [14]; unknown percent-level features in reactor spectra could matter, which the near detector TAO, whose data entered JUNO's 2026 analysis, is meant to address [11, 14]. NOvA projects that 3σ evidence for the normal ordering is achievable "within the next five years" over a range of plausible JUNO results, in a 2026 preprint [13].

Cosmology. Bound the sum of neutrino masses against the two floors [9]. Strongest evidence: the Bayes factor of 10 for the normal ordering under ΛCDM [9]. Main difficulty: the result depends on the cosmological model, and the data favour a sum below even the normal-ordering floor, which the authors say could hint at new physics "not necessarily related to neutrinos" [9].

Open sub-questions

  1. JUNO's local minima. How much more exposure does JUNO need to select the physical Δm²31 minimum under each ordering without the Daya Bay constraint [14]?
  2. Reactor spectrum features. Do percent-level features in the reactor spectrum, now measurable with the TAO near detector, change JUNO's preference for the normal ordering [11, 14]?
  3. The θ23 octant. How strongly does the true sin²θ23 control the joint Super-Kamiokande and T2K p-value for the inverted ordering, which ranged from 0.05 to 0.08 over the 90% intervals of θ23 and δCP [6]?
  4. T2K against NOvA. Is the better agreement of T2K and NOvA appearance data under the inverted ordering a statistical fluctuation [5, 8]?
  5. Below the floor. Does the preference of DESI and microwave background data for a mass sum below 0.059 eV persist in later DESI data, or vanish with evolving dark energy [9]?
  6. NOvA with JUNO. Which JUNO values of Δm²31 would leave NOvA short of 3σ for the normal ordering [13]?

Work on this in Papex

If you have a testable answer to one of the sub-questions above, you can draft it as a private hypothesis in Papex, or prepare the draft from Claude or ChatGPT with the Papex connector. You review the exact version, sign it and decide whether to publish it; a public record that names this problem is listed on this page. A Papex record shows who signed which version and when. It is not peer review and does not mean the hypothesis is correct.

Key references

  1. Qian, X. & Vogel, P. (2015). Neutrino mass hierarchy. Progress in Particle and Nuclear Physics 83, 1–30. 10.1016/j.ppnp.2015.05.002 · Checked 16 Oct 2026
  2. Abe, K., Abe, Ke., Aihara, H., Aimi, A., Akutsu, R., Andreopoulos, C. et al. (Hyper-Kamiokande Proto-Collaboration) (2018). Hyper-Kamiokande design report. Preprint, arXiv:1805.04163 (version 2, 28 November 2018), not peer reviewed. Source · Checked 16 Oct 2026
  3. Abi, B., Acciarri, R., Acero, M. A., Adamov, G., Adams, D., Adinolfi, M. et al. (DUNE Collaboration) (2020). Long-baseline neutrino oscillation physics potential of the DUNE experiment. The European Physical Journal C 80(10), 978. 10.1140/epjc/s10052-020-08456-z · Checked 16 Oct 2026
  4. Wester, T., Abe, K., Bronner, C., Hayato, Y., Hiraide, K., Hosokawa, K. et al. (Super-Kamiokande Collaboration) (2024). Atmospheric neutrino oscillation analysis with neutron tagging and an expanded fiducial volume in Super-Kamiokande I–V. Physical Review D 109(7), 072014. 10.1103/PhysRevD.109.072014 · Checked 16 Oct 2026
  5. Esteban, I., Gonzalez-Garcia, M. C., Maltoni, M., Martinez-Soler, I., Pinheiro, J. P. & Schwetz, T. (2024). NuFit-6.0: updated global analysis of three-flavor neutrino oscillations. Journal of High Energy Physics 2024(12), 216. 10.1007/JHEP12(2024)216 · Checked 16 Oct 2026
  6. Abe, K., Abe, S., Bronner, C., Hayato, Y., Hiraide, K., Hosokawa, K. et al. (Super-Kamiokande and T2K Collaborations) (2025). First joint oscillation analysis of Super-Kamiokande atmospheric and T2K accelerator neutrino data. Physical Review Letters 134(1), 011801. 10.1103/PhysRevLett.134.011801 · Checked 16 Oct 2026
  7. Capozzi, F., Giarè, W., Lisi, E., Marrone, A., Melchiorri, A. & Palazzo, A. (2025). Neutrino masses and mixing: entering the era of subpercent precision. Physical Review D 111(9), 093006. 10.1103/PhysRevD.111.093006 · Checked 16 Oct 2026
  8. NOvA Collaboration & T2K Collaboration (Abubakar, S., Acero, M. A., Acharya, B., Adamson, P., Anfimov, N., Antoshkin, A. et al.) (2025). Joint neutrino oscillation analysis from the T2K and NOvA experiments. Nature 646(8086), 818–824. 10.1038/s41586-025-09599-3 · Checked 16 Oct 2026
  9. Elbers, W., Aviles, A., Noriega, H. E., Chebat, D., Menegas, A., Frenk, C. S. et al. (2025). Constraints on neutrino physics from DESI DR2 BAO and DR1 full shape. Physical Review D 112(8), 083513. 10.1103/w9pk-xsk7 · Checked 16 Oct 2026
  10. Abubakar, S., Acero, M. A., Acharya, B., Adamson, P., Anfimov, N., Antoshkin, A. et al. (NOvA Collaboration) (2026). Precision measurement of neutrino oscillation parameters with 10 years of data from the NOvA experiment. Physical Review Letters 136(1), 011802. 10.1103/x53y-2b86 · Checked 16 Oct 2026
  11. Esteban, I., Gonzalez-Garcia, M. C., Maltoni, M., Martinez-Soler, I., Pinheiro, J. P. & Schwetz, T. (2026). Lessons from the first JUNO results. Journal of High Energy Physics 2026(4), 89. 10.1007/JHEP04(2026)089 · Checked 16 Oct 2026
  12. JUNO Collaboration (Abusleme, A., Adam, T., Adamowicz, K., Adey, D., Ahmad, S., Ahmed, R. et al.) (2026). Measurement of reactor neutrino oscillation with the first JUNO data. Nature 654(8118), 343–348. 10.1038/s41586-026-10538-z · Checked 16 Oct 2026
  13. NOvA Collaboration (2026). Determining neutrino mass ordering with NOvA and upcoming JUNO measurements. Preprint, arXiv:2606.14121 (version 1, 12 June 2026), not peer reviewed. Source · Checked 16 Oct 2026
  14. Adam, T., Ahola, T., Amanbek, N., An, F., Athayde Marcondes de André, J. P., Andreopoulos, C. et al. (JUNO Collaboration) (2026). Measurement of neutrino oscillation parameters at JUNO and indication of the neutrino mass ordering. Preprint, arXiv:2609.37895 (version 1, 29 September 2026), not peer reviewed. Source · Checked 16 Oct 2026

Page reviewed 16 Oct 2026

Source

Open, as listed by Wikipedia · Unsolved problems in physics. Checked 08 Oct 2026.

The status is the source’s. Papex does not decide whether a problem is solved.

Records on this problem

No public record addresses this problem yet.

More in Physics