Neutrino Atlas
EN
  1. Learn
  2. The three flavours

Chapter 1: The basics · Lesson 1.4

The three flavours

Confirmed

Neutrinos come in three types, each tied to one charged lepton: the electron, the muon and the tau. Three experiments spread over almost four decades established this pattern, and a precision measurement at CERN showed that no fourth light type couples to the Z boson.

Normal and inverted neutrino mass orderingTwo possible arrangements of the three neutrino masses. Each mass state is a mixture of the electron, muon and tau flavours.Normal orderingInverted orderingMass²ν1ν2ν3ν3ν1ν2Δm²₂₁|Δm²₃₁|Δm²₂₁|Δm²₃₁|νeνμντflavour content of each mass state
The two possible mass orderings. Each bar shows how much of each flavour a mass state contains (calculated from global-fit mixing angles: θ₁₂ ≈ 33°, θ₂₃ ≈ 49°, θ₁₃ ≈ 8.6°). Δm²₂₁ ≈ 7.4 × 10⁻⁵ eV² and |Δm²₃₁| ≈ 2.5 × 10⁻³ eV². Which ordering nature chose is still open; JUNO, DUNE and Hyper-Kamiokande aim to decide it. Spacings not to scale.
How an accelerator neutrino beam is madeProtons hit a target, magnetic horns focus the pions produced, the pions decay in a long tunnel into muons and muon neutrinos, an absorber stops everything except the neutrinos, which reach a near and a far detector.protonstargetmagnetic hornsdecay tunnelπ⁺ → μ⁺ + νμabsorbernear detectorfar detectorhundreds of kmExamples: T2K (J-PARC → Super-K, 295 km), NOvA (Fermilab → Ash River, 810 km), DUNE (Fermilab → SURF, 1,300 km)
The method goes back to the two-neutrino experiment at Brookhaven in 1962. Protons strike a target and produce pions, which magnetic horns focus into a beam. In the decay tunnel the pions turn into muons and muon neutrinos; an absorber of rock and steel stops everything else. Reversing the horn current selects negative pions and produces an antineutrino beam. The near detector measures the beam before oscillation, the far detector after it. Not to scale.

Beginner

One family name, three members

Physicists sort the smallest building blocks of matter into families. One family, the leptons, contains particles that do not feel the strong force that holds atomic nuclei together. The best-known lepton is the electron, the light particle that surrounds every atom. The electron has two heavier cousins: the muon, about 200 times heavier, and the tau, about 3500 times heavier. Both are unstable and decay within a tiny fraction of a second.

Each of these three charged particles has a neutral partner, a neutrino. Physicists call the three kinds of neutrino flavours, a playful word borrowed from ice cream. There is the electron neutrino, the muon neutrino and the tau neutrino. Together the six particles form three pairs, or generations.

How a neutrino shows its flavour

Neutrinos are almost impossible to see directly. They carry no electric charge and pass through the Earth with ease. On the rare occasions when a neutrino does hit an atom, it can turn into its charged partner. An electron neutrino produces an electron, a muon neutrino produces a muon, a tau neutrino produces a tau. The charged particle leaves a visible trail in a detector, and that trail reveals the flavour of the invisible neutrino that created it. It is a little like recognising an unseen car by the shape of its tyre tracks.

The second neutrino, 1962

Until 1962 most physicists assumed there was only one kind of neutrino. At Brookhaven National Laboratory near New York, Leon Lederman, Melvin Schwartz and Jack Steinberger built the first beam of neutrinos from a particle accelerator. The neutrinos came from the decay of short-lived particles called pions, which always produce a muon alongside the neutrino. When some of these neutrinos struck their detector, they produced muons and essentially never electrons. The neutrinos remembered that they had been born with a muon. This showed that the muon neutrino is a different particle from the electron neutrino1. The three physicists received the Nobel Prize in 1988.

The third neutrino, 2000

The tau was discovered in 1975, so a third neutrino was expected. Seeing it was hard, because a tau lives so briefly that it travels only about a millimetre before decaying. The DONUT experiment at Fermilab near Chicago used stacks of photographic film, so-called emulsion, that record particle tracks with a precision of a thousandth of a millimetre. In 2000 the team announced four events in which a short track ended in a tiny kink: a tau created by a tau neutrino, decaying almost at once2.

Why only three

Could there be a fourth or fifth neutrino? An answer came from CERN in Geneva. Its LEP collider produced millions of Z bosons, heavy particles that carry part of the weak force. A Z boson can decay into a neutrino and an antineutrino of any flavour. Each extra kind of light neutrino would give the Z one more way to disappear and would shorten its life. Measuring how long the Z lives, and therefore how spread out its mass appears, works like counting the drains in a bathtub by timing how fast it empties. The result is very close to three45.

This count applies only to light neutrinos that feel the weak force. Very heavy neutrinos, or so-called sterile neutrinos that ignore the weak force entirely, would not show up in it. Whether such particles exist is still an open question.

Flavours that change

One more surprise followed. The three flavours are not fixed labels. A neutrino born as one flavour can be detected as another after travelling a long distance. This effect, neutrino oscillation, only works because neutrinos have a tiny mass, and it is the subject of a separate article.

Quiz

Test your knowledge

4 questions on this level · Print as worksheet

  1. 1How much heavier than the electron is the muon?
  2. 2How does a detector reveal the flavour of an invisible neutrino?
  3. 3What did the 1962 Brookhaven neutrinos produce in the detector?
  4. 4How many light neutrino kinds did the Z boson measurements at LEP find?

Intermediate

Leptons in three generations

The Standard Model contains six leptons, arranged in three generations. Each generation pairs a charged lepton with a neutrino: (e, νₑ), (μ, νμ) and (τ, ντ). The charged leptons have very different masses, 0.511 MeV, 105.7 MeV and 1776.9 MeV6, but they interact in exactly the same way. Physicists call this lepton universality. The neutrinos are electrically neutral and interact only through the weak interaction (and gravity).

A neutrino's flavour is defined by the charged lepton that accompanies it in a charged-current interaction, mediated by a W boson. A muon neutrino striking a nucleus produces a muon; an electron neutrino produces an electron. Before neutrino oscillations were known, this was expressed as conservation of separate lepton flavour numbers Lₑ, Lμ and Lτ.

The two-neutrino experiment

In the late 1950s, the absence of the decay μ → e γ hinted that the neutrinos from muon decay might differ from those of nuclear beta decay. Bruno Pontecorvo and Melvin Schwartz independently proposed testing the idea with an accelerator neutrino beam. In 1962 a team led by Leon Lederman, Schwartz and Jack Steinberger directed 15 GeV protons from Brookhaven's Alternating Gradient Synchrotron onto a beryllium target. The resulting pions decayed in flight, π⁺ → μ⁺ νμ, and a 13.5 m steel wall stopped everything except the neutrinos. Behind it stood a 10-tonne aluminium spark chamber.

Over several months of running the chamber recorded a few dozen clean neutrino interactions. They produced long, penetrating muon tracks; the handful of shower-like events was consistent with background. Had electron and muon neutrinos been identical, electrons and muons would have appeared in comparable numbers. The neutrinos from pion decay were therefore a distinct particle, the muon neutrino1. The experiment also founded accelerator neutrino physics. Lederman, Schwartz and Steinberger shared the 1988 Nobel Prize in Physics.

Observing the tau neutrino

Martin Perl's group at SLAC discovered the tau lepton in 19757, which implied a third neutrino. Confirming it directly required a beam rich in ντ and a detector able to see a tau decay. The DONUT experiment (Fermilab E872) dumped 800 GeV protons from the Tevatron into a tungsten block. Charmed Dₛ mesons produced there decay occasionally to τ ντ, giving a small tau-neutrino component. The target was an emulsion cloud chamber: layers of nuclear emulsion interleaved with steel, read out after the 1997 run with automated microscopes.

A tau neutrino interaction produces a tau lepton that decays after roughly a millimetre. In the emulsion this appears as a short track followed by a kink. In July 2000 the collaboration announced four such events on an expected background of about 0.342. The final analysis in 2008 counted nine tau-neutrino events among 578 analysed neutrino interactions3. OPERA at Gran Sasso later saw muon neutrinos from CERN turn into tau neutrinos, confirming the third flavour through oscillation.

Counting light neutrinos at LEP

From 1989 the Large Electron–Positron collider at CERN produced Z bosons at rest. Scanning the beam energy across the Z peak traces its line shape: a resonance whose width ΓZ reflects the total decay rate. Part of that rate is invisible, Z → νν̄, and every light neutrino flavour adds the same amount. Comparing the invisible width with the width into charged leptons gives the number of light neutrinos:

Nν = (Γinv / Γ_ℓ) · (Γ_ℓ / Γν)SMΓinv: measured invisible width of the Z; Γ_ℓ: measured width into one charged-lepton pair; (Γ_ℓ/Γν)SM: Standard Model ratio, about 1/1.991.

The four LEP experiments ALEPH, DELPHI, L3 and OPAL already found Nν ≈ 3 in 1989. Their combined final result was Nν = 2.9840 ± 0.00824. A 2020 re-evaluation of the luminosity measurement moved this to 2.9963 ± 0.00745, fully consistent with three. Neutrinos heavier than half the Z mass, about 45 GeV, or neutrinos without weak couplings, are not counted.

Quiz

Test your knowledge

4 questions on this level · Print as worksheet

  1. 1What is the mass of the tau lepton?
  2. 2What stopped everything except neutrinos in the 1962 Brookhaven experiment?
  3. 3What signature identified a tau neutrino in DONUT's emulsion?
  4. 4What value of Nν resulted from the 2020 re-evaluation of the LEP luminosity?

Expert

Flavour as a weak-interaction label

In the Standard Model the left-handed leptons form SU(2)L doublets Lα = (να, ℓ_α)L with α = e, μ, τ. The flavour index is defined by the charged current

J^μCC = Σα ν̄α γ^μ (1 − γ⁵) ℓ_αThe W boson couples each flavour neutrino να to its own charged lepton ℓ_α with universal strength g/(2√2).

Flavour eigenstates are therefore defined only through interactions. Since neutrinos are massive, they are superpositions of mass eigenstates, |να⟩ = Σᵢ U*_αi |νᵢ⟩, with U the PMNS matrix. The separate lepton flavour numbers are violated by oscillations, though charged-lepton flavour violation remains unobserved. The charged-lepton masses are mₑ = 0.51099895 MeV, mμ = 105.6583755 MeV and mτ = 1776.93 ± 0.09 MeV6. Direct kinematic limits on the flavour-state masses are weak outside the electron sector: mνμ < 0.19 MeV (90 % CL, π⁺ decay) and mντ < 18.2 MeV (95 % CL, τ decays at LEP)6.

Brookhaven 1962

Danby, Gaillard, Goulianos, Lederman, Mistry, Schwartz and Steinberger used 15 GeV protons from the AGS on a beryllium target, a decay region for the secondary pions and a 13.5 m steel shield in front of a 10-tonne spark chamber of ninety aluminium plates, each about 2.5 cm thick1. Most of the flux came from π → μν_μ with some K decays. The chamber recorded a few dozen single-muon events above cosmic background, and only a handful of shower candidates compatible with background. The expected ratio of electron to muon showers for a single neutrino type would have been of order one. The result established νμ ≠ νₑ and explained the non-observation of μ → eγ at the then expected rate. The 1988 Nobel Prize recognised both the discovery and the neutrino-beam method.

DONUT and the direct observation of ν_τ

After the discovery of the τ in 19757 and of its weak isospin partner through τ decay kinematics at LEP, the ντ was firmly expected. Fermilab E872 (DONUT) produced a prompt beam by stopping 800 GeV Tevatron protons in a tungsten dump; a few per cent of the neutrino flux at the detector was expected to be ντ, mainly from Dₛ → τντ and the subsequent τ decays. Magnets and passive shielding removed charged particles and most muons. The target consisted of emulsion cloud chamber modules with 1 mm steel plates and emulsion sheets, followed by a scintillating-fibre tracker and a muon spectrometer that located the interaction vertex for emulsion scanning.

The signature was a charged track of length ~1 mm with a measurable kink, with no muon or electron identified at the primary vertex. Four candidates on an expected background of 0.34 ± 0.05 were reported in 20002. The final 2008 analysis of 578 located interactions gave nine ντ CC events over a small estimated background, and a cross-section measurement consistent with lepton universality within large errors3. DONUT thus completed the direct observation of all three Standard Model neutrinos.

The Z invisible width

At the Z pole the partial width into one neutrino pair is

Γν = GF MZ³ / (12√2 π) · ρ ≈ 167 MeVGF: Fermi constant; MZ ≈ 91.19 GeV; ρ: electroweak correction close to 1. Each light flavour that couples to the Z contributes Γν to Γinv.

LEP measured MZ = 91.1875 ± 0.0021 GeV, ΓZ = 2.4952 ± 0.0023 GeV and Γinv = 499.0 ± 1.5 MeV from the hadronic and leptonic line shapes4. The ratio form Nν = (Γinv/Γ_ℓ)(Γ_ℓ/Γν)SM cancels most theoretical and luminosity uncertainties, with (Γν/Γ_ℓ)SM = 1.991 ± 0.001. The LEP–SLD combination gave 2.9840 ± 0.0082, about two standard deviations below three. Janot and Jadach showed in 2020 that an improved Bhabha cross-section calculation and a correction for beam–beam effects on the luminosity measurement shift the value to 2.9963 ± 0.00745, which PDG adopts6. Direct counting in e⁺e⁻ → νν̄γ (single-photon events) gives Nν = 2.92 ± 0.05, independent but less precise6.

What the count constrains

  • Only states with m < MZ/2 and standard couplings to the Z are counted. Heavy neutral leptons and gauge-singlet sterile neutrinos evade it; mixing with active states would reduce Γinv slightly.
  • A fourth sequential generation would need a neutrino heavier than ~45 GeV; together with Higgs production rates at the LHC this scenario is excluded.
  • Cosmology measures a related but different quantity, the effective number of relativistic species Neff (Standard Model value 3.044), which also counts any light particle that was thermally populated in the early universe.
  • Oscillation data require exactly three mass splittings only within the three-flavour framework; global fits such as NuFIT 6.0 find a consistent picture with |Δm²₃₁| ≈ 2.5 × 10⁻³ eV² and Δm²₂₁ ≈ 7.5 × 10⁻⁵ eV²8.

The three-generation structure itself, including why the charged-lepton masses span nearly four orders of magnitude while neutrino masses are at least six orders smaller than the electron mass, has no explanation within the Standard Model.

Quiz

Test your knowledge

4 questions on this level · Print as worksheet

  1. 1What is the Z partial width into one neutrino pair?
  2. 2Why is Nν extracted with the ratio form (Γinv/Γ_ℓ)(Γ_ℓ/Γν)SM?
  3. 3What direct kinematic limit applies to the tau-neutrino mass?
  4. 4How does the cosmological Neff differ from the LEP neutrino count?

Open questions

Why there are exactly three generations is unknown. Whether additional neutral leptons exist, either light sterile neutrinos or heavy partners linked to the seesaw mechanism, is being tested at short-baseline experiments, colliders and in cosmology.

Terms to explore

People

On the timeline

  1. 1962Brookhaven experiment discovers the muon neutrino
  2. 1975Tau lepton discovery implies a third neutrino
  3. 1988Nobel Prize for the muon neutrino
  4. 1989LEP shows there are three light neutrinos
  5. 2000DONUT observes the tau neutrino
  6. 2015OPERA establishes tau-neutrino appearance

References

  1. G. Danby et al., Observation of high-energy neutrino reactions and the existence of two kinds of neutrinos, Phys. Rev. Lett. 9, 36 (1962), doi:10.1103/PhysRevLett.9.36
  2. K. Kodama et al. (DONUT Collaboration), Observation of tau neutrino interactions, Phys. Lett. B 504, 218 (2001), doi:10.1016/S0370-2693(01)00307-0
  3. K. Kodama et al. (DONuT Collaboration), Final tau-neutrino results from the DONuT experiment, Phys. Rev. D 78, 052002 (2008), doi:10.1103/PhysRevD.78.052002
  4. ALEPH, DELPHI, L3, OPAL, SLD Collaborations, LEP Electroweak Working Group, SLD Electroweak and Heavy Flavour Groups, Precision electroweak measurements on the Z resonance, Phys. Rept. 427, 257 (2006), doi:10.1016/j.physrep.2005.12.006
  5. P. Janot, S. Jadach, Improved Bhabha cross section at LEP and the number of light neutrino species, Phys. Lett. B 803, 135319 (2020), doi:10.1016/j.physletb.2020.135319
  6. S. Navas et al. (Particle Data Group), Review of Particle Physics, Phys. Rev. D 110, 030001 (2024), doi:10.1103/PhysRevD.110.030001
  7. M. L. Perl et al., Evidence for anomalous lepton production in e+e− annihilation, Phys. Rev. Lett. 35, 1489 (1975), doi:10.1103/PhysRevLett.35.1489
  8. I. Esteban, M. C. Gonzalez-Garcia, M. Maltoni, I. Martinez-Soler, J. P. Pinheiro, T. Schwetz, NuFit-6.0: updated global analysis of three-flavor neutrino oscillations, JHEP 12, 216 (2024), doi:10.1007/JHEP12(2024)216

Last updated: · Editorial standard

Cite this page

Text

Neutrino Atlas (2026). “The three flavours”. Retrieved 22 November 2026, from https://neutrino-atlas.com/learn/the-three-flavours/

BibTeX

@misc{neutrinoatlas2026learnthethreeflavours,
  title        = {The three flavours},
  author       = {{Neutrino Atlas}},
  year         = {2026},
  url          = {https://neutrino-atlas.com/learn/the-three-flavours/},
  note         = {Accessed 2026-11-22},
  language     = {english}
}