THE evidence for the polarization of the light from many stars has accumulated since Hiltner's first observation of the effect1. Owing to the correlation of the effect with interstellar absorbing medium, it now seems certain that the explanation must be found in an anisotropy of this medium. The direction of the polarization is variable from star to star, shows a regional correlation, and a general preference for the electric vector to lie in the galactic plane. The magnitude of the effect (up to 12 per cent plane polarization) would be accounted for by a 5 per cent difference in the extinction coefficient in the two planes of oscillation2.