When Hall (1949) and Hiltner (1949) began searching for intrinsic polarization in distant OB stars, they found that a clearly coherent pattern of position angles was present across the sky in several directions in the Milky Way. The pattern could not be due to the intrinsic polarization of the individual stars, since their rotation axes were likely to be randomly distributed over the sky. Some other, global, mechanism must be responsible for the strong correlation in position angle from star to star. The interstellar origin of this phenomenon is not in doubt, as only stars suffering interstellar extinction by dust are affected and there is a positive correlation between polarization and extinction (reddening). The accepted model for interstellar polarization is dichroism resulting from the presence of asymmetric grains that are aligned with the galactic magnetic field (cf. Whittet 2003). If the direction of alignment changes along the line of sight, the interstellar medium (ISM) also exhibits birefringence, producing a component that is circularly polarized. Studies of interstellar linear and circular polarization are important because they provide information both on grain properties (size, shape, refractive index) and on the galactic magnetic field. How are we sure that it is the magnetic field that determines the alignment direction of interstellar dust? In the very diffuse ISM, the position angles of the polarization agree with the position angles from observations of synchrotron radiation in the radio (with an offset of ~90°, expected for zero pitch-angle electrons, c.f. Jackson 1975). For example (Fig. 9.1), both synchrotron observations (Sukumar and Allen 1991) and near infrared imaging polarimetry (Jones 1989b) of the external galaxy NGC 4565 show the magnetic field is parallel to the dust lane. These two very different physical mechanisms reveal the same magnetic field geometry. The identity of the alignment mechanism has proved to be an intriguing problem in grain dynamics that has teased theorists for many years (see the review by Voshchinnikov 2012 and Chapter 6).