when the waveform descriptor index is zero then the return has no waveform and the offset value can be disregarded. when the waveform descriptor index is non-zero then the return is pointing to *some* waveform that can in theory be located anywhere in the file and whose size and sampling is described by the referenced waveform descriptor and whose location is described by the offset. In theory (unlikely to happen except for LAS files with just 1 - 5 returns) there could be just one waveform in the WDP file that is shared by all returns. More likely is that for each shot there is one complete waveforms and all returns point to this same waveform. But each waveform can also be replicated once for every return. Either in its entirety or only the 32 digitization samples before and after it. So there is not one definite answer.