One of the observers noted that when following a source for an extended period of time the source drifts in the order of 10 pixels (1.5 arcsec) per hour. Examination of the data showed that the drift is mostly in the vertical direction with respect to the telescope. The fact that the drift is similar in V and I excludes differential diffraction affects, while the straight tracks for the drifts makes it unlikely to be a problem in the tracking or guiding. The most likely cause is flexure of the camera with respect to the 45 degree pick-off mirror. Checking the mirror fixation itself does not show any slack, but it might be that the whole probe moves.