2004A&A...421..215M

We use a set of carefully selected, published average multifrequency polarimetric observations for six bright cone-dominated pulsars and devise a method to combine the multifrequency polarization position angle (PPA) sweep traverses. We demonstrate that the PPA traverse is in excellent agreement with the rotating vector model over this broad frequency range, confirming that radio emission emanates from perfectly dipolar field lines. Correcting for the effect of retardation we firmly establish the steepest gradient point in the combined PPA traverse to be the fiducial phase in these pulsars. We use this combined curve and inputs from earlier studies to determine the geometrical angles of the neutron star, namely α - the angle between the rotation and the dipole magnetic axis and β - the angle between the magnetic axis and the observers line of sight. Using these estimates of α and β we derive the geometrical emission heights (r_geo). Further, using the relativistic beaming model based on effects of aberration and retardation, we find the delay emission heights r_delay^BCW suggested by Blaskiewicz et al. (1991ApJ...370..643B). We find in general r_delay^BCW<r_geo, which can be explained by a broad emission region operating in pulsars or/and the signature of the magnetic field sweepback effect as suggested by Dyks et al. (2004ApJ...607..939D). For pulsars with central core emission in our sample, we find the peak of the central core component to lag behind the steepest gradient of the PPA traverse at several frequencies. Significant frequency evolution of the core width is observed over this frequency range. The above facts strongly suggest: (a) the peak core emission does not lie on the fiducial plane containing the dipole magnetic axis and the rotation axis, and (b) the core emission does not originate from the polar cap surface.