The ultimate goal of conventional seismic data processing is to derive a response that one can assume is similar to the zero-offset compressional-wave reflection response of the earth. Using recorded multifold seismic data, traces that have different source-receiver offsets and that illuminate the same source-receiver midpoint (or reflecting point) are summed. That is done by applying NMO correction to different offset traces and then stacking them. The mean reflection amplitudes are displayed at their zero-offset time, and the display is assumed to be similar to the zero-offset section. Such an approach works when the reflection amplitudes across different offsets do not change. However, we know that is usually not the case, and in fact reflection amplitudes do change with offset. The change occurs because energy partitioning between reflected and transmitted wave modes is angle-dependent. The P-wave reflected and transmitted amplitudes depend on the compressional-wave velocities and the shear-wave velocities, as well as on the density of the incident and reflecting media.