Okay, it is an interaction between POS and REP...? The higher the REP the further through the sample each repetition of P1-P3 advances. I think. And the playback position resets with the repetition cycle. Still not sure what controls/sets the original sample start position. Possibly it is set by the position of POS and advances by the length of LEN. That would almost make sense. Not sure yet.
Sorry for live blogging my fairly inane and surely confused thoughts. I figure it might be useful to someone else in the future to read through this nonsense.
If anyone understands this better than I do or has an alternate theory or knows the programming intimately, please correct the record.