2015MNRAS.454.2216D

The known population of pulsars contains objects with four- and five-component profiles, for which the peak-to-peak separations between the inner and outer components can be measured. These Q- and M-type profiles can be interpreted as a result of sightline cut through a nested-cone beam, or through a set of azimuthal fan beams. We show that the ratio RW of the components' separations provides a useful measure of the beam shape, which is mostly independent of parameters that determine the beam scale and complicate interpretation of simpler profiles. In particular, the method does not depend on the emission altitude and the dipole tilt distribution. The different structures of the radio beam imply manifestly different statistical distributions of RW, with the conal model being several orders of magnitude less consistent with data than the fan-beam model. To bring the conal model into consistency with data, strong effects of observational selection need to be called for, with 80 per cent of Q and M profiles assumed to be undetected because of intrinsic blending effects. It is concluded that the statistical properties of Q and M profiles are more consistent with the fan-shaped beams, than with the traditional nested-cone geometry.