Database structure

[JBirdieH]

We have an 3 paradox applications. Server NT, clients NT & 95. One of those apps crashes/locks repeatedly. I am now must accept the file structure is not suitable. We are trying to enter completely dissimilar legal documents for the primary purpose of lookup on the name associated (for liens/divorces/lawsuits/etc.) In UNIX the table was simply 15 fields, one of which was a name field. This did little for name search capacity. (The name relationship is generally always one to many with the document). We could not come up with a suitable alternative when we moved to paradox, so now there are two tables, one with the 15 fields of document indexing (1-11 primary key), and second table with the 15 fields of document indexing AND a first and lastname fields. The two tables are referentially linked and the secondary indexes are larger than the .db file. We didn't want to go with autoincrementing or another meaningless key. Does anyone have suggestions on a structure that would perform better? Note also that the documents entered are so dissimilar, the contents of each field vary with the document type. Data has been rearranged to keep number types, date types, etc in the same field, but the meaning of that data is different for each document. There is not a finite set of document types.