Help w/ data modeling, 1 to 1, 1 to none? hack?

[Ron Creamer]

Greetings,

I need advice on the optimal way to spec my sql database tables. I'm stuck between a couple of lousy alternatives to a rather simple problem which I'm sure has been solved before.

Let's assume there is a table called 'paper'
Each paper record holds all the characteristics of this type of paper.
Some papers, but not nearly all, have a unique characteristic. In my example, the unique characteristic is that this paper is "FSC Certified".

For those that are FSC Certified, there will be a 2 fields to track. My thinking is to normalize by putting these two fields in another table, let's call it the 'fsc' table. The two fields are simply a varchar field holding the FSC Certificate Number, and the other field will be a blob/scan/PDF of the actual certificate.

So in my 'paper' table, I'm thinking that I should have a fk (foreign key) to the 'fsc' table's pk (primary key).

But this doesn't work because only about 15% of my 'paper' table records are FSC Certified. And the other 85% would have a null foreign key.

Any advice on how to better model this?

Thanks in advance for your help.