An aluminum sphere the size of a large globe sits in a neutron beam and answers each hit with a precise note. Inside the sphere there is nothing but nitrogen and eleven tiny steel balls, each the size of a pinhead. When a neutron enters the sphere, it knocks a particle out of a nitrogen atom, and that jolt turns into an electrical signal. That is how the detector works, and it has just been calibrated for the first time on neutrons with a single, strictly defined energy.
Neutrons leave no trace in matter directly: they have no charge. To catch them, you need a middleman, and nitrogen turns out to be a handy interpreter. A neutron hits a nitrogen nucleus, and out comes either a proton or an alpha particle. These charged fragments can then be seen: they ionize the gas, and each free electron is amplified at the anode, like an avalanche running down a slope. The more energy the neutron has, the stronger the signal, and from its height you can tell the particle's energy.
Earlier attempts with such nitrogen spheres used continuous spectra, where neutrons have a range of energies. But no one had calibrated them on a monoenergetic beam, where every particle arrives with a known energy. In Athens, a tandem accelerator produced neutrons at seven different energies, from 0.75 to 2.75 MeV, and the detector's response was measured for each. It turned out that the reconstructed energy grows exactly along a straight line with the beam energy. For the reaction with alpha emission, the slope of that line is one, with an uncertainty of only one and a half percent.
Such a linear response means the sphere can be used not just as a counter that says "a neutron was here," but as a spectrometer that says "a neutron was here with an energy of exactly 1.8 MeV." This matters for calibrating other detectors and for monitoring neutron fields in accelerator halls. The sphere hears every hit and answers with a precise note, like a tuning fork set to the neutron's voice.
The detector only reached stable gain 18 hours after the high voltage was switched on. The reason for such a long plateau time is not yet clear, but after that, the gain remained stable.