This is a small problem for us too. Within the 3T data, there are actually two non-unique datasets subject 24528, as Andrew noted, and subject 68050. If it would not be too difficult, the data will be more flexibly dealt with if the subject ID's were unique across sites. Because the numbers are not contiguous within sites, and the great majority of files are unique, we assumed the numbering to be constructed to give unique ID numbers.

Subject sub24528 is not unique. Found in /Volumes/elrond/share/data/fcon_1000/alldata/3.0T/NewYork_a_subjects.txt /Volumes/elrond/share/data/fcon_1000/alldata/3.0T/Oxford_subjects.txt
Subject sub68050 is not unique. Found in /Volumes/elrond/share/data/fcon_1000/alldata/3.0T/Bangor_subjects.txt /Volumes/elrond/share/data/fcon_1000/alldata/3.0T/Leiden_2200_subjects.txt
2 subjects found in multiple sites