I realized tonight that there was a problem with the observatory computer; the program that keeps the computer’s internal clock synchronized with Internet time servers wasn’t working.
It turns out the problem had nothing to do with that program. The IP addresses of campus DNS servers had changed so DNS lookup was failing.
When DNS was reconfigured the synchronization program reset the computer’s internal clock, adjusting it by -46.089 seconds. In other words, 46.089 seconds was subtracted from the computer’s internal clock.
Prior to tonight (9/16/2013) the last synchronization with the time server was on 8/20/2013 at about 5:45AM.
The bottom line is that the timestamp on images taken on the dates below are likely all off by some amount between 0 and -46 sec, but there is no way to know (at least at the moment) how big that offset is. On any given night the time difference between images should be very accurate. It is only the “absolute” time that is incorrect.