> OK, then the bad table may be one of the user accessible system tables.
> I'd try select count (*) from RDB$RELATIONS, the RDB$RELATION_FIELDS,
> the RDB$FIELDS ... and continue until you find the bad table. I'd like
> to know which it was...

It was the table SYS$DMLCACHE. I use this table for the DML Caching
mechanism of IBO.

Regards

Guido