We have studied the radio pulsar emission beam assuming a) a magnetic dipole field geometry and b) that the beam geometry is defined by the field lines that are not contained within the light cylinder. In general, the beam is compressed in the meridional direction. When the magnetic and rotation axes are aligned the beam is circular and as, the angle between these axes increases, the ratio of meridional to longitudinal dimension decreases monotonically to the minimum value 0.62 when the axes are orthogonal. This beam shape is thus consistent with that inferred from the study of circular polarization in average pulse profiles by Radhakrishnan and Rankin. Evidence for meridional compression is also found in the extensive observational study of Lyne and Manchester (1988).The beam evolution was determined using this data set, the beamwidth being found proportional to P–1/2, where P is the pulsar period. This relation implies that the more rapidly rotating pulsars should have larger beams, and this should aid in their detection. The more numerous, slower pulsars should have somewhat smaller beams than previously determined. This implies that the pulsar birthrate is probably close to the highest current estimates (1 in 25yr).