>OK, what am I missing here?
>
>I have two files: BALANCE and PTYPES
>
>PTYPES is ordered on the basis of its TYPE field
>
>BALANCE is related to PTYPES through the TYPE field
>
>Every TYPE in BALANCE should have a match in PTYPES.
>
>I deliberately alter a TYPE in BALANCE to ensure that at least one record does not have a match.
>
>If I SET FILTER TO EMPTY(PTYPES.TYPE) I duly find the erroneous record.
>
>If I SELECT BALANCE.* FROM BALANCE,PTYPES WHERE BALANCE.TYPE=PTYPES.TYPE AND NOT EMPTY(PTYPES.TYPE) INTO DBF ANYTABLE
>
>it duly selects all the correct records and omits the faulty one.
>
>Yet, if I try to do the opposite (in order to detect the faulty one)
>i.e.
>
>SELECT BALANCE.* FROM BALANCE,PTYPES WHERE BALANCE.TYPE=PTYPES.TYPE AND EMPTY(PTYPES.TYPE) INTO DBF ANYTABLE
>
>I get zilch.
>
>Pourquoi?
>
>Harry

Do you have buffering enabled? You have to commit the change you made before you can see it in a SQL statement. Just a SWAG.