Puzzling Pulsars and Supernova Remnants

The fact that the majority of the youngest radio pulsars are surrounded by expanding supernova remnants is strong evidence that neutron stars are produced in the supernovae of massive stars. In many cases, the pulsar appears significantly offset from the geometric centre of the supernova remnant, indicating that the neutron star has moved away from the site of the explosion with a substantial space velocity since birth. Here we show that the these offsets show an overwhelming preference for one sign in terms of Galactic longitude, a result that has important implications for the number of genuine associations. The origin of this statistically significant effect may lie in a differential Galactic rotational velocity between stars and gas in the interstellar medium. An outstanding question in pulsar statistics is what fraction of proposed pulsarsupernova remnant associations are genuine as opposed to chance line-of-sight alignments. In attempting to answer this question, Gaensler & Johnston (1995) noted that the distribution of positional offsets shows a clear excess at small values compared to that expected by chance. These analyses suggest that between 7 and 17 of the 30 presently proposed pairs are genuine (Lorimer et al. 1998). Resolving the offsets into Galactic coordinates we find an important and unexpected twist in this story. The positional data for 27 pulsar-supernova remnant pairings presented in Fig. 1 are the result of a cross-correlation of the publicly-available pulsar and supernova remnant catalogues (Taylor et al. 1995; Green 1998). For each pairing we computed the positional offset defined as the difference between the position of the pulsar and the geometric centre of the remnant. Following Shull et al. (1989) we normalised each offset to the angular size of the respective remnant. Although pulsar positions are known to within an arc second from timing measurements, positions of remnant centroids are more difficult to determine. We carefully searched the available literature to check the positions listed in the catalogue and obtain conservative estimates of their uncertainty. It is immediately apparent that the distribution of offsets shown in Fig. 1 is not uniform. This is easily quantified by counting the number of pairs N within a given angular radius r. For a uniform distribution, N(< r) ∝ r2 so that the number expected within 2r should be 4N . This is clearly not the case, with 10 and 18 pairs lying within half and one remnant radii respectively. What is most remarkable is the sign of the offsets. Whilst the signs in Galactic latitude