Sort command not working as expected

view story

http://unix.stackexchange.com – I have got a dataset like this [email protected],2009-11-27 [email protected],2009-11-27 [email protected],2009-11-27 When I am running command to delete all of the same entries of column2 sort -t ',' -k2 stars.txt -u It is deleting the entry of column1. In order to delete the duplicate entries of column2, I am having to enter -k3 flag sort -t ',' -k3 stars.txt -u Can anyone explain to me why it is happening? Why I have to enter +1 to the column in the file to delete the column? (HowTos)