On 8 June 1998, the Super-Kamiokande Collaboration announced at the Neutrino 1998 conference in Takayama, Japan, that atmospheric muon neutrinos disappear over long baselines. The data, published in Physical Review Letters later that year, was the long-sought proof that neutrinos change flavor as they travel — and therefore that they have mass. The Standard Model needed its first major extension in fifty years.
Takaaki Kajita’s presentation at Takayama is widely seen as the discovery moment. He won the 2015 Nobel Prize for it.
What came before
For thirty years, atmospheric and solar neutrino experiments had reported “anomalies” — fewer events than predicted. The atmospheric anomaly came from Kamiokande, IMB, and Soudan-2, all observing that the ratio of muon-flavor to electron-flavor neutrinos from cosmic-ray-induced atmospheric showers was below expectation. Several explanations were possible:
- Wrong flux predictions (cosmic-ray spectra, hadronic interactions)
- Detector systematics (mis-identification of event types)
- Neutrino oscillation (muon-flavor neutrinos converting to tau before detection)
Distinguishing these required more data and better systematics control. Super-Kamiokande, with 50,000 tonnes of ultrapure water (vs. Kamiokande’s 3,000 tonnes) and 11,000 photomultiplier tubes, was built to do exactly this.
The setup
Super-Kamiokande is a water Cherenkov detector in the Kamioka mine, 1,000 m underground. The 50-kt cylindrical tank is lined with 11,146 inward-facing PMTs that detect Cherenkov light from charged secondary particles produced when neutrinos interact in the water.
Atmospheric neutrinos arrive from all directions — cosmic rays produce them in every patch of upper atmosphere. The key physics:
- Downward-going neutrinos travel ~20 km (the atmospheric thickness) before reaching Super-K.
- Upward-going neutrinos have crossed the Earth, traveling up to ~13,000 km.
- If oscillation is occurring, the longer baseline gives more time for flavor change.
The Super-K collaboration counted muon-like and electron-like events as a function of arrival direction, and looked for the asymmetry pattern predicted by oscillation.
The data
The 1998 paper presents data from 535 days of running, with about 5,000 fully-contained single-ring events in the relevant energy range.
The two key plots:
-
Zenith-angle distribution of muon-like events: Downward-going events match prediction without oscillation. Upward-going events show a clear deficit — about 50 % of the expected rate. Electron-like events at the same energies and angles show no deficit — consistent with not oscillating to electron flavor over the relevant baseline.
-
L/E distribution: When events are binned by the ratio of baseline to energy, the muon-event rate as a function of L/E shows the characteristic oscillatory dip — fewer events at higher L/E.
The two patterns together rule out flux uncertainties or detector systematics: both are flavor-symmetric and direction-symmetric, while the data show a clear flavor- and direction-dependent effect.
The fit
The collaboration fit the data to two-flavor oscillation between muon-neutrino and tau-neutrino. Best-fit parameters:
- $\sin^2(2\theta_{atm}) > 0.82$ at 90 % CL
- $\Delta m^2_{atm} \approx 2.2 \times 10^{-3}$ eV² (with broad allowed range $5 \times 10^{-4}$ to $5 \times 10^{-3}$ eV²)
The implication: the difference between two neutrino masses is non-zero. Therefore at least one neutrino has a non-zero mass. The Standard Model’s assumption of massless neutrinos was wrong.
Why the discovery was decisive
Several features of the analysis made it convincing:
Multiple independent observables agreed. The zenith-angle distribution, the L/E distribution, the flavor ratio, the energy dependence — all consistent with the same oscillation parameters.
Backgrounds were small. The deep underground location, ample event statistics, and good flavor reconstruction meant the signal-to-background was strong.
The flux uncertainty cancelled. Because the analysis compared downward (control) to upward (signal) events of the same type and energy, atmospheric-flux uncertainties cancelled to leading order.
Other experiments later confirmed. SNO (2001) for solar neutrinos, KamLAND (2002) for reactor antineutrinos, T2K and NOvA (later) for accelerator beams. Each independent test landed in the same oscillation framework.
Why it took 30 years from anomaly to discovery
The 30-year gap between the original atmospheric anomalies of the 1960s–1980s and Super-K’s 1998 announcement is instructive about how physics discoveries actually happen:
- The original anomalies had large statistical and systematic uncertainties.
- Flux predictions improved over decades (Honda et al. atmospheric flux models)
- Detector technologies improved (Kamiokande → Super-K → ten-times statistics)
- Theoretical predictions matured (MSW effect, full three-flavor framework)
- The pattern of consistent results across many independent experiments built confidence
No single experiment “proved” oscillation in isolation. The 1998 Super-K paper was decisive because it provided the cleanest single dataset — but it was confirming a hypothesis the community had been considering seriously for two decades.
What came next
The 1998 paper was the start of the oscillation era, not its end. In the seven years that followed:
- 1999: SNO begins data taking (heavy water for all-flavor solar detection)
- 2001: SNO publishes its first measurement of all-flavor solar neutrino flux — proves solar oscillation too.
- 2002: SNO publishes the all-flavor / electron-flavor ratio — direct proof of solar electron-to-other-flavor oscillation.
- 2002: KamLAND sees energy-dependent reactor antineutrino oscillation
- 2004: Super-K publishes the full three-flavor analysis with L/E shape
- 2012: Daya Bay and RENO measure $\theta_{13}$ at 5σ
By 2010, the oscillation framework was as well-tested as any extension of the Standard Model.
How to read the paper
The original paper — Y. Fukuda et al., PRL 81, 1562 (1998) — is 5 pages of dense text plus references. The structure:
- Introduction: 2 paragraphs on neutrino oscillation theory and previous results
- Detector and analysis: 1 page on Super-K and event selection
- Results: 2 pages of plots and tables
- Discussion and conclusion: 1 page on the fit and significance
The paper assumes the reader knows two-flavor oscillation phenomenology. The key plots are Figure 1 (zenith-angle distributions for muon-like and electron-like events) and Figure 2 (the L/E plot). Look at those first.
For context, the longer 2005 follow-up paper by the same collaboration (Ashie et al., PRL 93, 101801) provides much richer detail and is also worth reading.
What to read next
- Super-Kamiokande 1998 in depth — our long-form article
- The complete guide to neutrino oscillation — full theory + history
- Takaaki Kajita — the 2015 Nobel laureate’s biography
- Detector deep dive: Super-Kamiokande — how the detector actually works
- The 2015 Nobel Prize: Kajita and McDonald — the prize coverage
For the original 1998 paper, search “Evidence for oscillation of atmospheric neutrinos” — it is available open-access on the PRL site and on arXiv:hep-ex/9807003.
Frequently asked
What paper does this spotlight cover?
Y. Fukuda et al. (the Super-Kamiokande Collaboration), 'Evidence for oscillation of atmospheric neutrinos,' Physical Review Letters 81, 1562–1567 (1998). The eight-page paper that announced the discovery of neutrino oscillation.
Why was this paper so important?
It was the first direct experimental proof that neutrinos change flavor in flight — and therefore that they have mass. Both consequences were major surprises. The Standard Model assumed massless neutrinos; oscillation forced its first concrete extension in fifty years. Takaaki Kajita won the 2015 Nobel Prize for this work.
How could a single experiment prove oscillation?
By measuring the angle-dependence of muon-neutrino events. Atmospheric neutrinos arrive from all directions: cosmic rays produce them in the upper atmosphere everywhere on Earth. Neutrinos from directly overhead travel ~20 km before reaching the detector; neutrinos from straight below travel ~13,000 km through Earth. Super-K found that the upward-going (long-baseline) muon-neutrino rate was about half of what would be expected without oscillation, while the downward (short-baseline) rate was as expected. The asymmetry pattern matched oscillation almost exactly.
What were the measured parameters?
The 1998 result implied $\sin^2(2\theta_{atm}) > 0.82$ and $\Delta m^2_{atm} \sim (5{-}50) \times 10^{-4}$ eV². With twenty more years of data, these have been refined to $\sin^2\theta_{23} \approx 0.57$ and $|\Delta m^2_{31}| \approx 2.51 \times 10^{-3}$ eV² — within the original range.
Cite this article 5 formats
APA
Explainers Desk. (2025, August 17). Paper Spotlight: Super-Kamiokande 1998 — atmospheric neutrino oscillation. Neutrino Times. https://neutrino-times.com/articles/paper-spotlight-super-k-1998/
Chicago
Explainers Desk. "Paper Spotlight: Super-Kamiokande 1998 — atmospheric neutrino oscillation." Neutrino Times, August 17, 2025. https://neutrino-times.com/articles/paper-spotlight-super-k-1998/.
MLA
Explainers Desk. "Paper Spotlight: Super-Kamiokande 1998 — atmospheric neutrino oscillation." Neutrino Times, 17 Aug. 2025, https://neutrino-times.com/articles/paper-spotlight-super-k-1998/.
BibTeX
@misc{neutrino-times-paper-spotlight-super-k-1998,
author = {Explainers Desk},
title = {Paper Spotlight: Super-Kamiokande 1998 — atmospheric neutrino oscillation},
howpublished = {Neutrino Times},
year = {2025},
month = {aug},
url = {https://neutrino-times.com/articles/paper-spotlight-super-k-1998/},
note = {Accessed: 2025-08-17}
} RIS
TY - GEN TI - Paper Spotlight: Super-Kamiokande 1998 — atmospheric neutrino oscillation AU - Explainers Desk PY - 2025 DA - 2025-08-17 PB - Neutrino Times UR - https://neutrino-times.com/articles/paper-spotlight-super-k-1998/ ER -