A simple phys. picture which correctly predicts the flip angle effect in Fourier transform spectroscopy discussed by R. R. Ernst et al. (1974) is presented. The approach is based on populations alone and does not require knowledge of the d. matrix formalism. Nonselective pulses are represented by a cascade of consecutive pulses which affect only certain sets of transition at a time. The relative timing of the pulses within the cascade is unimportant provided the intervals are all short with respect to nuclear precession rates and relaxation. [on SciFinder (R)]