JB>then join the 255 lists together again into one list and

JB>repeat using the next least significant byte,
JB>repeat unitl all bytes of the key value have been used,

JB>the list is now sorted.

Ahhh... Radix sort now comes back to me.

JB>AFAIK there's nothing more efficient than O(n) sorting.
JB>but remember you do need memory for the linked list

JB>it's possible with a few tweaks do radix sort on IEEE floating point
JB>numbers too, but it'll never work for strings etc...

It seems to me that if Radix sort works correctly with multi-byte 
integers and floats, it should work with strings.

Why won't it?

þ CMPQwk 1.42 999

--- Maximus/2 3.01