>>91
This is now correct.
'(0 1 2 3 4)
You might be interested in the iota procedure.
(member S R)
This is the part that holds back the speed. If you look at your final S, you will find that it grows roughly like the square root of R. But for each S jump test you are searching the entire R storage from the highest value downwards. You can tell exactly which R value will be hit by the next S jump and avoid searching entirely.