A 95- X 95- X 95-cm nuclear grade graphite stack was bombarded with a well-collimated monoenergetic 14.75- ± 0.05-MeV fast neutron beam from a tritium target of a neutron generator. The neutron spectra measured in such types of integral experiments are susceptible to the various neutron interactions (elastic and inelastic scattering by the first few excited levels including anisotropy of angular distributions). This, in turn, facilitates identification and treatment of discrepancies between the experimental and calcula-tional results. The neutron spectra were measured with a 50- X 50-mm NE-213 liquid scintillator using the pulse shape discrimination technique to reject gamma-ray counts. The linearity test of the neutron spectrometer was performed by means of radioactive gamma-ray sources and D(d,n)He3 and T(d,n)He4 neutrons. Amplification factors (in light units per channel) were achieved with a 11Na22 radioactive source. The spectrometer was checked with the D(d,n)He3, T(d,n)He4 reactions and an americium-beryllium radioactive neutron source. The measured proton recoil spectra were unfolded in the neutron spectra by the FORIST unfolding code.