The Pakistan rules for 2002 have the transition at 0:01 instead of 0:00. This disagrees with the commentary: # From Rives McDow (2002-03-13): # I have been advised that Pakistan has decided to adopt dst on a # TRIAL basis for one year, starting 00:01 local time on April 7, 2002 # and ending at 00:01 local time October 6, 2002. This is what I was # told, but I believe that the actual time of change may be 00:00; the # 00:01 was to make it clear which day it was on. The remaining rules in the file are at 0:00. Also, our friends at timeanddate.com have it at midnight as well. http://www.timeanddate.com/time/change/pakistan/karachi?year=2002 Of course, the problem with any fall-back transition after 0:00 and before 1:00 is that it implies the calendar date goes forward, then back, then forward again. In this case going from Oct 6, back to Oct 5, then forward to Oct 6 again. While there are indeed areas where this certainly occurred, we should be careful not to introduce that artificially. If there is uncertainty between 0:00 and 0:01, we should assume 0:00 (IMHO). Thanks, Matt