By the end of 2010 we found a mysterious behaviour in long dark exposures, where both the count level and structure was depending on the internal focus of the instrument, repeatable in several measurements spread over a month. This behaviour disappeared after the instrument was opened in January 2011. It is not at all clear what has changed. We are monitoring the long darks at different internal foci every now and then, but the reported effect from 2010 has not been reproduced after February 2011.