Magnetic mirage in reflection of light from imperfect medium

I argue that the separation of the optical response into nonreciprocal (magnetic) and reciprocal parts can be unequivocally done in a microscope designed to image magnetic structure, only when the source and the detector satisfy the condition of the exact backscattering. Usually the condition is not under control, and the magneto-optic setup may display a magnetic mirage---a spurious picture of a random magnet. The contrast of the speckle-type mirage created by imperfections in the object increases together with the spatial resolution. The pseudomagnetic rotation angle is estimated.