DONUT: catching a tau before it disappears
The third neutrino flavor was expected long before its interactions were directly identified. The difficult clue was a tiny track that changed direction when a short-lived tau decayed.
Expected is not the same as directly identified
The discovery of the tau lepton in 1975 suggested a third charged-lepton family, alongside the electron and muon. Its decays and other measurements supported the existence of an associated neutrino. But that was not yet the same experiment as identifying a tau neutrino interacting in a target.
An electron-neutrino interaction can produce an electron; a muon-neutrino interaction can produce a muon. To complete the analogous test for tau flavor, the experiment had to recognize a tau produced at the primary interaction.
That requirement made the task difficult. The tau is heavy, so producing it requires sufficient energy. It is also short-lived. Instead of a long penetrating trail, the detector must often recognize a small segment followed by decay products.
Build a beam with a small tau-neutrino component
DONUT used an energetic proton beam at Fermilab, stopped in a tungsten target. The collisions produced short-lived particles, including charmed mesons, whose decays supplied neutrinos. Decays involving a tau provided a source of tau neutrinos and antineutrinos.
The incoming sample also contained electron and muon flavors. A candidate could therefore not be identified merely by saying it came from the beam. The experiment needed information from the outgoing tracks and from the interaction vertex itself.
Magnets and thick shielding reduced the other particles reaching the target. Electronic detectors helped locate interactions and identify charged particles. Nuclear emulsion then supplied the fine spatial detail needed to find a short decay within that much coarser initial location.
- 01Primary vertex
Locate the interaction and identify its particles
- 02Short decay
Resolve a path followed by a kink
- 03Check alternatives
Charm decays and secondary interactions
The evidence was a short path and a change of direction
Nuclear emulsion is a photographic material in which charged particles leave microscopic trails. Layer after layer can be scanned and aligned to reconstruct tracks in three dimensions. Its detail comes with substantial work: the interesting vertex must be found among many unrelated segments.
A tau can travel a short distance and decay into a charged daughter particle moving in a different direction. The two segments form a kink. The relevant signature combines the path length, angle, momentum information and the other particles at the original vertex.
A kink alone is insufficient. A charged particle scattering from material can also turn. A charmed particle produced in an electron- or muon-neutrino interaction can decay nearby. Missing the original electron or muon could make that event resemble a tau-neutrino interaction.
Four candidates, compared with an explicit background
The first paper analyzed 203 located neutrino interactions and selected four events meeting its tau-decay criteria. The estimated total background from charm decays and secondary hadronic interactions was 0.34 events. Those numbers refer to the selected analysis, not to every trigger recorded by the apparatus.
The important comparison was not “four is a large number.” It was that four events with these properties were difficult to obtain from the specified background model. Particle identification and event selection made that comparison possible; their limitations belonged in the calculation.
The target exposures took place in 1997. Fermilab announced the result in July 2000, the preprint appeared that December, and the journal paper followed in 2001. Keeping those dates separate makes clear how much reconstruction and analysis came after collecting the tracks.
Direct evidence still arrives through an inference
“Direct observation” does not mean the neutrino itself left a visible line. It means the experiment identified its characteristic charged-current interaction through a tau and its decay. The unseen incoming particle was inferred from a measured final state and tested alternatives.
The DONUT collaboration brought together institutions in the United States, Japan, Korea and Greece. The result depended on beam operation, shielding, electronic tracking, emulsion development, automated scanning and careful reconstruction. The tiny track was supported by a large measurement chain.
A later final analysis expanded the event sample and studied the interaction cross section. The original four-event result retains its own historical role: it supplied the direct tau-flavor interaction evidence that earlier family counting and decay measurements had anticipated.
Primary sources & revision
- Perl et al. · Evidence for Anomalous Lepton Production (1975)
- DONUT Collaboration · Observation of Tau Neutrino Interactions (2000/2001)
- Fermilab · First Direct Evidence for Tau Neutrino (July 2000)
- DONuT Collaboration · A first measurement of the tau-neutrino cross section (2007/2008)
- Fermilab · How an experiment fixed a hole in the Standard Model (2025)
First published and source-checked on 9 October 2026. Original explanatory text and diagrams by Sandbox Physics. Illustrations are schematic; no experimental event records are reproduced here. This is an independent educational publication, not an official Nobel or experiment collaboration publication.