I know so little about relational databases that I'm worried about being misled by an unreliable guide found on the internet. Reading suspiciously, this guide seems garbled.<p>Section 3, second and third normal forms, says "Under second and third normal forms, a non-key field must provide a fact about the key, us the whole key, and nothing but the key. In addition, the record must satisfy first normal form."<p>There seems to be a stray word "us". Ignoring that, this is cute word play that doesn't quite make sense. If your table has non-key fields you are inevitably providing facts about the non-key fields.<p>Continuing,<p>"We deal now only with "single-valued" facts. The fact could be a one-to-many relationship, such as the department of an employee, or a one-to-one relationship, such as the spouse of an employee."<p>This is the wrong way round. Usually there are lots of employees and a few deparments. Each employee works for just one department, but each department has many employees. Thus the "department of an employee" is a many-to-one relation, or function, which takes an employee and yields a department. The one-to-many relationship here is the employee list of a department.<p>The example for 3.2 seems to be opening the wrong can of worms. Suppose that Mr Strauss, who works for the department of waltz in Vienna, is seconded to the department of piety in Rome, in order to teach them some dance steps. Then we want his row in the database to read<p>(Straus, Waltz, Rome)<p>So one can of worms is sticking generic labels on your fields. If you label your fields (Employee, Department, Employee-location) there is no problem. If you label your fields (Employee, Department, Department-location) you have a problem, but it is obvious. If you label your fields (Employee, Department, Location) you are heading for trouble as some users of the database fill in the location of department and other users of the database fill in the location of the employee.<p>Hmm, second and third normal form are suspiciously similar, differing only because we regard some fields as belonging to the key. Is the article trustworthy?