A necessary correction for relating local network magnitude scales to Richter's local magnitude (ML) involves accounting for the shape of the far-field body-wave spectrum of the phases used for determining magnitude. When not corrected for, this effect causes errors of about one magnitude unit at ML ∼ 3 for some southern California earthquakes. The discrepancy should be comparable for ML > 3, but at smaller magnitudes will decrease with decreasing ML. It may be corrected for either by direct comparison of network scales with magnitudes determined from Wood-Anderson seismograms, or by spectrum measurements over a range of magnitudes. The nature of the discrepancy and the corrections required to account for it are demonstrated by an example, the aftershocks of the 1968 Borrego Mountain, California earthquake.