The thermal spectrum of a beryllium-moderated intermediate assembly is calculated in the approximation of a heavy crystalline moderator. The results are compared to experiment, and the qualitative agreement is found to be good. In the particular case calculated, the temperature dependence of the spectrum is so weak that a straightforward Fermi age calculation is also in good qualitative agreement. The most important uncertainties in the calculation are due to the heterogeneous distribution of neutron absorbing material, and the use of the heavy mass approximation in the presence of strong absorption.