>Very interesting bruce. I will start testing that myself and I'll be looking forward to hear what you find.

Here's the latest: I loaded small test data, and no problems. I did a larger real load today, and the problem returned. So, I'm going to clear it all out again (keeping an extra empty DB structure this time for future use), and add new data bit by bit, testing along the way, to see if I can isolate it better.

I'm fairly sure it has something to do with indexes, just like that old FPW ISO problem, except it's more complex now...