Using a refractive scintillation event of PSR 1237+25, Wolszczan & Cordes (1987) inferred a projected displacement of the emission region of order 108 cm across the pulse. Gupta (these proceedings) inferred a lower separation limit of a few times 107 cm from a similar event for PSR 1133+16. In both cases the projected displacement was also a nonmonotonic function of pulse phase.

Neither these large separations nor the nonmonotonic behavior can be explained in a simple model which assumes that the emission originates parallel to the field lines of a magnetic dipole, at a fixed altitude across the open field zone, and propagates in a straight line to the observer. There is considerable circumstantial evidence for dipolar fields from polarization sweep and pulse morphological studies. Time delays due to different emission altitudes could produce asymmetries in pulse profiles and projected displacements, but cannot explain the large separations if the emission originates well within the light cylinder.