>what are your field types, what is the compound key expression and what is
>the critiria you wanted for uniqueness (because evidetly they don't match..)
>
>Arnon
>
>
>
>>
>>Problem:
>>In a table with a compound 'candidate' index my code checks the contents
>for an existing match to the compund key, using a seek().
>>If no record is found, then it inserts a new record.
>>Much to my surprise I have found up to 20% of all records are duplicated
>records.
>>
>>Any suggestions on avoiding this problem?
>>Thanks in advance,
>>Art Castro

Thanks for your interest in helping.

The fields are Field1 C(12) and Field2 C(16)
and the index is Field1 + Field2.

The seek uses Memvar1 + memva2. Both memvars are previously converted with a padr() function to be the correct length for each.

The table accepts the insertion of the duplicated value without an errror message.

When I use and SQL the duplictes show up. Same as with a broewse with 'nooptimze'. If I don't use 'nooptimize' in the browse the duplicates don't show.