You can certainly INSERT or UPDATE a field with an empty string. Someone put it in there. It may have been something like an automated process that imported data from an external source, and imported empty strings as empty strings rather than checking them to see if they were empty and converting them to NULL.

I don't think that John's assertion that empty strings shouldn't exist is entirely accurate. NULL means that the value does not exist. An empty string means that it does exist, but is blank. I'm trying to come up with a good reason for a difference, and am coming up short. But here's a somewhat questionable example:

Say you're entering people's names into a database, first, middle, and last. You might not have some people's middle names available when you enter the data, so you'd leave them as NULL, meaning that the data does not exist. Later, when completing the data entry, you put in people's middle names, replacing the NULL. But then you come across someone who doesn't have a middle name at all. You might then want to change that field to the empty string, indicating that the value does exist, but it is empty. Of course, all of that is a logical construct. You could conceivably design it to be the other way around. It's all about how you interpret your data. (Though the way I describe makes more sense to me.)

So it's possible that the data you found that had empty strings may have had a different meaning than the data that had NULL values. Or it may have been a data entry error. Only a definition of what the database fields mean for your situtaion can really let us know.