Daya Bay: why compare a near detector with a far one?

Before calling something missing, know what should arrive

Suppose a detector records fewer reactor antineutrino interactions than a calculation predicts. Have some antineutrinos changed flavor, or did the calculation overestimate the reactor output? A small shortfall is hard to interpret when the absolute source prediction has a comparable uncertainty.

Daya Bay’s answer was to measure at different distances. Detectors near the reactor cores pinned down the emitted signal; detectors farther away tested whether it changed as expected over the longer journey.

The experiment was built in Guangdong, China, by an international collaboration. Its 2012 discovery analysis used six reactor cores and six antineutrino detectors spread over two near halls and one far hall. This article discusses that configuration, not the later stages of the experiment.

Sources: [1] · [2]

The far hall should see fewer events even without oscillation

Particles spread out as they travel away from a compact source. A distant detector therefore receives a smaller flux even if no flavor change occurs. Simply dividing the raw far count by the raw near count would confuse ordinary geometric dilution with oscillation.

The prediction has to account for each detector’s distance from each reactor, its target size, its operating time and its efficiency. With several reactor cores, the near and far halls also receive different mixtures of source contributions.

Once those differences are included, the near measurements provide a reference for the expected far signal. The question becomes: is the remaining far deficit larger than can be explained by the uncertainties of that comparison?

Sources: [1]

What cancels and what does not

Consider an error that makes the predicted yield of every reactor too high by the same factor. It changes the absolute predictions at both near and far sites. A comparison anchored by the near data is much less sensitive to that shared normalization error.

Now suppose one detector misses slightly more delayed neutron signals than its neighbour. That relative difference in efficiency can mimic part of a near–far deficit, and taking a ratio does not remove it.

Daya Bay used functionally identical detectors, calibrated their responses and studied backgrounds such as accidental coincidences and particles produced by cosmic-ray muons. The first paper reported a combined uncorrelated detection uncertainty of 0.2 percent. Precision came from measuring the differences that did not cancel.

Sources: [1] · [2]

A six-percent deficit became a mixing measurement

In the 55 days of data behind the first result, the far hall’s rate after background subtraction was about 94 percent of what the near detectors predicted. After correcting for exposure and geometry, that is a deficit of roughly six percent.

In a three-neutrino oscillation analysis, the result established a nonzero value of the mixing angle called theta one-three at 5.2 standard deviations. The equation gives the reported mixing amplitude, with the statistical and systematic uncertainties separated.

The six-percent rate deficit and the mixing amplitude are different quantities. The deficit is averaged over reactor distances, energies and detector response; the amplitude is a parameter inferred by fitting the oscillation model to the data.

Sources: [1]

The result opened the way for other experiments

A nonzero theta one-three made electron-flavor appearance in long-baseline accelerator experiments a more promising route to studying the three-flavor system. It helped establish the conditions for investigating differences between neutrino and antineutrino oscillations.

Daya Bay’s reactor disappearance result did not itself discover CP violation or explain why the universe contains more matter than antimatter. Those questions need other measurements and further reasoning. Daya Bay supplied one precise ingredient, not the whole programme.

The method works beyond neutrinos too. If the two instruments and their surroundings are well understood, a second detector can turn a hard absolute prediction into a better-controlled comparison. In the reactor Lab, finding candidate pairs is the first step; Daya Bay went on to compare the resulting samples with far greater precision.

Sources: [2] · [1] · [3]

Try it in the Lab

Primary sources & revision

1. Daya Bay Collaboration · Observation of electron-antineutrino disappearance (2012)
2. Daya Bay / Brookhaven · First results announced, 8 March 2012
3. Particle Data Group · Neutrino Masses, Mixing, and Oscillations (2024)

First published 2026-10-09; last revised 2026-10-12. 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.