Mass Ordering: Normal vs. Inverted Hierarchies

The Standard Model of particle physics, after the discovery of neutrino oscillation, contains three light neutrinos. Each has a definite mass — though they are not the flavor states ${\nu }_e$, ${\nu }_{\mu }$, ${\nu }_{\tau }$ that interact in detectors but rather mass eigenstates conventionally labeled ${\nu }_1$, ${\nu }_2$, ${\nu }_3$. The three flavor states are linear combinations of these three mass eigenstates, related by the unitary PMNS mixing matrix.

Oscillation experiments measure the differences between mass-squared values: $\Delta m_{21}^2\equiv m_2^2-m_1^2\approx 7.42\times 10^{-5}{\text{ eV}}^2\text{ (solar splitting)}$ $|\Delta m_{31}^2|\equiv |m_3^2-m_1^2|\approx 2.51\times 10^{-3}{\text{ eV}}^2\text{ (atmospheric splitting)}$

The solar splitting is known including its sign, because solar matter effects (MSW) require ${\nu }_2$ to be heavier than ${\nu }_1$. The atmospheric splitting, however, is known only in magnitude. The sign — whether ${\nu }_3$ is heavier than ${\nu }_1$ or lighter — is the mass ordering question, and it remains one of the major open questions in neutrino physics.

This post explains the two possibilities, why the ordering matters, and how it will be measured.

The two scenarios

If $\Delta m_{31}^2>0$, the mass ordering is normal: $m_1<m_2<m_3$. The smallest splitting (solar) lies between the two lightest states, and the larger splitting separates this pair from the heaviest state.

If $\Delta m_{31}^2<0$, the ordering is inverted: $m_3<m_1<m_2$. The solar splitting again lies between ${\nu }_1$ and ${\nu }_2$, but now both are above the lightest state ${\nu }_3$.

The flavor composition of each mass eigenstate is the same in both scenarios (set by the PMNS matrix elements), but the assignment of “lightest” versus “heaviest” changes. Some literature uses “hierarchy” instead of “ordering”; the modern preference is “ordering” because the term “hierarchy” implies a large mass ratio that may not be present (the absolute mass scale is small enough that all three states could be close in absolute mass).

Why the ordering matters

Several aspects of fundamental physics depend on which ordering nature realizes.

Neutrinoless double beta decay. The effective Majorana mass parameter relevant for $0\nu \beta \beta$ is: $⟨m_{\beta \beta }⟩=\left|{\sum }_i{U}_{ei}^2{m}_i\right|$ For inverted ordering, $⟨m_{\beta \beta }⟩$ has a non-zero lower bound around 15 meV regardless of the unknown CP phases. For normal ordering, cancellations among the three terms allow $⟨m_{\beta \beta }⟩$ to drop to nearly zero in a “funnel” region of parameter space. So if the ordering is inverted, $0\nu \beta \beta$ should be detectable by the upcoming next-generation experiments (LEGEND-1000, nEXO). If normal, detection is possible but not guaranteed.

Cosmological constraints on $\sum m_{\nu }$. The minimum sum of masses in inverted ordering is approximately 0.10 eV; in normal ordering, it is approximately 0.06 eV. Recent cosmological bounds (CMB + BAO) give $\sum m_{\nu }<0.12$ eV at 95% C.L., increasingly squeezing the inverted-ordering prediction. If the cosmological bounds tighten further to around 0.08 eV without detection, inverted ordering would be in tension with cosmology.

Theoretical model preferences. Many neutrino-mass models (especially those with discrete flavor symmetries) make different predictions for the two orderings. A measured ordering would discriminate between competing models.

Long-baseline experiments. The CP-violating phase ${\delta }_{CP}$ in the PMNS matrix is measured by comparing ${\nu }_{\mu }\to {\nu }_e$ and ${\bar{\nu }}_{\mu }\to {\bar{\nu }}_e$ appearance rates. Matter effects in the Earth, which depend on the ordering, partially mimic CP violation. Disentangling matter effects from genuine CP violation requires knowing the ordering.

How the ordering is measured

Three experimental approaches are converging.

JUNO at Jiangmen, China — operational from 2026. JUNO is a 20-kiloton liquid-scintillator detector observing reactor antineutrinos at a baseline of 53 km. At this baseline the survival probability has both a fast oscillation (driven by $\Delta m_{31}^2$) modulated by a slower oscillation (driven by $\Delta m_{21}^2$). The two interfere differently in the two orderings, producing a distinctive spectral signature.

JUNO’s planned sensitivity, with a six-year exposure, is approximately 4σ for distinguishing the two orderings — possibly approaching 5σ if systematic uncertainties are kept under control. The first results are expected by 2030.

NOvA and DUNE use long-baseline accelerator neutrinos. A muon-neutrino beam from Fermilab or J-PARC propagates through the Earth’s mantle to a far detector. The matter potential modifies the oscillation pattern differently for the two orderings:

For neutrinos with normal ordering, the appearance probability $P({\nu }_{\mu }\to {\nu }_e)$ is enhanced by the matter effect; for inverted ordering, it is suppressed. For antineutrinos, the effects reverse. Comparing neutrino and antineutrino appearance rates therefore probes the ordering directly.

NOvA (Fermilab to Soudan, 810 km) has been running since 2014 and has reported a slight preference for normal ordering, currently around 1-2σ. DUNE (Fermilab to South Dakota, 1300 km) — beginning operation around 2030 — will have larger matter effects and reach approximately 5σ ordering significance within the first few years.

Atmospheric neutrinos at IceCube and Super-Kamiokande’s atmospheric program also probe the ordering through matter effects in the Earth. The integrated atmospheric flux passing through Earth at different zenith angles shows ordering-dependent oscillation features. PINGU (a proposed IceCube core upgrade) and ORCA (a KM3NeT-affiliated project) are designed to extend this approach with better resolution at the relevant energies.

Current status

As of 2026, individual analyses give the following preferences (rough):

• NOvA + T2K combined: weak preference for normal ordering, approximately 1-2σ.
• IceCube atmospheric: weak preference, similar level.
• Cosmological bounds: weak tension with inverted ordering at the upper edge (depending on prior assumptions).

No single experiment has reached the 5σ discovery threshold. Combined analyses give marginally tighter constraints but still below discovery.

The forthcoming results from JUNO and DUNE should resolve the question. By approximately 2030-2032, the combined experimental data should give a definitive answer with greater than 5σ confidence.

What if it’s normal?

If the ordering turns out to be normal:

• The minimum $\sum m_{\nu }$ is around 0.058 eV.
• $⟨m_{\beta \beta }⟩$ can be anywhere from approximately 1 meV (cancellation funnel) to 5 meV (maximally far from the funnel).
• Cosmological bounds remain comfortably consistent.
• $0\nu \beta \beta$ may or may not be detectable in upcoming experiments — the funnel region is hard to reach.

What if it’s inverted?

If the ordering turns out to be inverted:

• The minimum $\sum m_{\nu }$ is around 0.097 eV — close to current cosmological bounds.
• $⟨m_{\beta \beta }⟩$ has a definite lower bound around 15-20 meV.
• $0\nu \beta \beta$ should be detectable by KamLAND-Zen, LEGEND, nEXO if Majorana neutrinos exist at all.
• Theoretical models predicting inverted ordering would be supported.

A long-standing open question, soon to close

The mass-ordering question has been open since the early 2000s, when the existence of three distinct neutrino masses was firmly established. For two decades, experiments have edged closer to determining the sign of $\Delta m_{31}^2$ but have not crossed the discovery threshold.

The remaining wait is short. JUNO operations begin in 2026; NOvA continues to accumulate data; DUNE comes online around 2030. By the early 2030s, the combined data set should give a clean answer.

The answer matters. It will affect the strategy for the next round of $0\nu \beta \beta$ experiments, the interpretation of cosmological data, and the structure of the leptonic CP-violation programme. It is one of the few remaining “binary” answers the field has ever waited for at this kind of patience.

Within a few years, we will know — for the first time — whether the heaviest neutrino is the one most strongly tied to the muon-tau flavor sector (${\nu }_3$), or one of the two more electron-flavor-mixed states. Either answer reshapes the picture.

In the meantime, both orderings remain in play. The experimental programme is converging. The waiting is almost over.