Cowan and Reines: finding the neutrino in two flashes

Look for what the neutrino leaves behind

A nuclear reactor produces a great many electron antineutrinos, the antimatter counterparts of electron neutrinos. Almost all pass through nearby matter. Occasionally, one interacts with a proton and produces two other particles: a positron and a neutron.

Those two particles gave Clyde Cowan, Frederick Reines and their colleagues a way to recognize the interaction. A positron is the electron’s positively charged counterpart. It slows down and annihilates with an electron, producing gamma rays. The neutron takes longer: it slows down and is eventually captured by a nucleus, producing another signal.

The experiment reported in 1956 used water targets containing cadmium, surrounded by scintillation detectors that produced light when radiation deposited energy in them. The apparatus at Savannah River was built to catch the products of the reaction. These were the visible clues to a particle that had arrived unseen.

Sources: [1] · [2]

The useful pause between the signals

The first signal arrives promptly. The second follows when the neutron is captured, in this apparatus by cadmium, and releases gamma radiation. A pair arriving in the right order and close together in time is more distinctive than either signal alone.

Researchers still have to decide how long to wait for the second signal. Wait too briefly and some genuine neutron captures will arrive after the window has closed. Wait much longer and unrelated pulses have more chances to be mistaken for the partner.

The equation below shows this tradeoff in a simple model. Neutron-capture delays follow an exponential distribution with a chosen mean waiting time. A wider window accepts more genuine pairs, but the gain gradually levels off.

Sources: [2]

Sometimes two unrelated signals line up

Detectors also record pulses from ordinary radiation and other sources. Suppose one pulse happens to look like the first signal and an unrelated pulse soon follows it. The pair can satisfy the timing rule even though no neutrino interaction produced it.

These are called accidental coincidences. They become more common when either kind of single pulse becomes more frequent, or when the waiting window gets longer. The estimate below expresses that simple idea.

One way to estimate this background is to shift one set of recorded times far enough to break the true short-delay pairs, then count the new pairs. The accidental overlaps remain. In our Lab, this shifted record gives a control sample that you can compare with the original selection.

Sources: [1] · [2]

How the team checked the explanation

The historical experiment did more than find close pairs. The researchers examined the signal energies, the waiting times, and what changed when the reactor was operating. They also investigated other radiation that could imitate the expected signal.

Each check answered a different doubt. Were the energies appropriate for the reaction? Were the delays consistent with neutron capture? Did the signal depend on the reactor? Could reactor-related neutrons or gamma rays account for it instead?

Together, these comparisons made the antineutrino interpretation convincing. The 1956 paper is worth reading with those questions in mind: much of the discovery lies in how carefully the team tested the other explanations.

Sources: [1] · [2]

Try making the window wider

Open the reactor Lab and acquire a record. Keep that record fixed while you widen the coincidence window. Watch how many genuine pairs are accepted, and how many accidental pairs come along with them. The total count can rise even when the extra events make the sample less useful.

Next switch off the simulated reactor and acquire a new record. Unrelated background pulses still produce coincidences. Finally, raise the background rate and compare the recorded pairs with the shifted control. The control is an estimate, so its count will fluctuate too.

This exercise uses illustrative rates and a simple capture-time model. It assumes the pulse categories are known and leaves out effects such as missed pulses and detector dead time. It teaches one part of the historical method with simulated data; the original 1956 records are not used.

Sources: [2]

Try it in the Lab

Primary sources & revision

1. Cowan, Reines, Harrison, Kruse & McGuire · Detection of the Free Neutrino (1956)
2. Reines et al. · Detection of the Free Antineutrino (1960)

First published 2026-10-09; last revised 2026-10-09. 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.