Hooray! The code Dirk supplied gave me a clean load with just under 4500 revisions. Should I proceed with this dataset or wait for additional changes from Dirk and Stephen's on-going discussion? Scott