Experiments in a large zero-power graphite reactor indicate that simple-point reactor kinetics adequately describe the neutron fluctuation spectra everywhere except near the edges of the reactor. Near the edges, the break frequencies and rolloff slopes differ significantly, and anomalous irregularities in the distribution are observed. Computations explicitly for the large reactor, based on recently proposed theories, also agree excellently away from the edges, and fail to predict the measured spectra near the boundary.