Variations of count rate with orientation of the detector were investigated. Scaled difference images (SDI) were generated from pairs of arrays acquired at different detector orientations. The hypothesis tested was that any variations could be totally explained by random disintegration noise. We accepted or rejected this hypothesis according to the statistical significance of the difference between the measured and theoretical means and variances of the SDIs. When two images acquired at the same angle were compared, no significant differences were noted. Differences were significant between most pairs of images obtained at different angular orientations. Cameras with inadequate magnetic shielding were more sensitive to orientation effects than those with proper magnetic shielding. The mode of variation strongly suggests an interaction with a magnetic field.