How quickly do you want it?

I guess it's just the definitions, examples of how it changes something (before and after) and example of what goes wrong if you don't do it. No high-falutin' theory or any of that junk, and nothing controversial? Certainly no mention of NULLs because that would start a war.

Incidentally, I think I saw something a few weeks ago - an article on SQL Server Central (or perhaps written by a threadizen but posted somewhere else) that covered 1NF through 3NF very well.

If covering normal forms 1 through 5 WITHOUT the intermediates like EKNF and BCNF because the article can get too long, and leaving out 6NF (other than to say that it is applicable to schemas that have to handle time-dependent data, and avoids some of the problems that people often run into with that) because it is too hard for this sort of article, and leaving out DKNF (because most things don't have a DKNF form, there's no way to normalise them to DKNF) would be OK, I can do it if it's not urgent. But I'm heading back to the UK next Wednesday so wouldn't be able to start on it before about May 7th. Also, I'm not sure that covering all of forms 1 to 5 in a single article is going to produce something short enough, it might be better to split it in half with 1NF, 2NF and 3NF in one article; 4NF, 5NF and the non-technical one or two liner on 6NF in another.

Incidentally, the 6NF mentioned above is a Date invention and I regard it as a very good thing despite disagreeing with him on a couple of other things (MVLs and NULL, mostly). The other thing that used to be called 6NF is what is now called DKNF and is pretty useless (its uselessness is something else that Date and I have a common view on). Don't get the impression from the stuff going on in the normalisation topic in the relational theory forum that I think Date is wrong about everything, or that David Portas thinks Codd was wrong about everything.