A very brief statement of the problems of nuclear reactor design prefaces an account of some of the nuclear-physics measurements necessary. The time-of-fiight neutron spectrometer installed at the Atomic Energy Research Establishment is described in detail with reference to these measurements. The performance of the machine, which makes use of a 3.2-MeV linear electron accelerator as neutron source, is shown in comparison with that previously published for other spectrometers. It is shown that the energy resolution obtainable at energies over 10 eV with the A.E.R.E. spectrometer is rather better than has previously been available for comparable counting rates. This is illustrated by comparison of experimental observations of the cobalt total neutron cross-section in the neighbourhood of a scattering resonance at 120 eV.