Shortly after orthogonal modes of polarization were observed in pulsar radio emission (Manchester et al. 1975), a series of theoretical papers (Cocke & Pacholczyk 1976, Melrose & Stoneham 1977, Melrose 1979) were quick to point out that these abrupt, 90 degree transitions in polarization position angle could be explained by a propagation effect. More specifically, the plasma above pulsar polar caps may be birefringent. The orthogonal modes separate because they have different indices of refraction. So, as Allen & Melrose (1982) point out, the plasma is somewhat analogous to a terrestrial birefringent crystal, at least in the way it affects the ray paths of the modes. With very sensitive observations of individual pulse polarization, Stinebring et al. (1984) concluded that orthogonal modes are superposed, meaning they occur simultaneously. This observation is consistent with the birefringence hypothesis. Stinebring et al. also suggested that pulsars depolarize at high radio frequency via the superposed modes. Barnard & Arons (1986) expanded on the earlier theoretical work, and proposed that, in addition to orthogonal modes, birefringence could account for the broadening of pulsar average profiles at low radio frequency and their depolarization at high frequency. The purpose of this paper is to further investigate the suggestions of Stinebring et al. and Barnard & Arons.