>Ok. This is a simplified versio of my previous question, because I have simplified the problem
>
>I have two datetime fields in a table.
>field1, field2
>I have indexes on each of them.
>I need all of the records for which field1>field2
>So I index on field1>field2
>
>SELE * from Mytable where field1>field2
>
>gives no Rushmore optimization. Why not?

Rushmore only works when one side of the expression EXACTLY matches the index expression. Neither field1 or field2 match your index expression of field1>field2. If you have an index on field1 or on field2 then your expression would be optimizable. To optimize your index you would need something like (field1>field2) = .T.