I was hoping that you'd say that. I agree. All parts of the address are known is it is known that the 2nd address line has a value of nothing rather than being unknown which is what a NULL would indicate.

In an even more practical sense, the use of NULL's for "end dates" can also be even more bothersome and I'll use the proverbial "end of time" date (9999-12-31) instead of a NULL.

The problem with both of those is, or course, duplication of "non data" similar to the "conventional" use of NULL. Some folks could imply that such duplication in those columns I speak of is justification for additional normalization and that may even be true. It just wouldn't be practical in my humble opinion. That's where I thought you guys were going with all of this and I'm relieved that's not the case... Heh... I didn't want to be a renegade for yet another thing in SQL. 🙂