Except PKS 1830-211, all sources detected in our observations are extended. Both their morphologies and spectral indices suggest that sources , , and are previously unnoticed SNRs. Sources and could be remnants with particularly flat spectra as well. The most intense of these galactic sources is , with an integrated flux density of 17.7 Jy at 1.42 GHz. None known radio pulsar is associated with this object. The surface brightness is W m-2 Hz-1 sr-1, and its distance, according to a relationship (Milne 1979), would be roughly of 2.5 kpc. This implies a radius of 32.5 pc. In a standard ISM ( cm-3) and assuming a Sedov phase for the expanding shell, the age of the SNR would be yr.
In Fig. 2 (click here) we have superimposed to the filtered radio map the confidence contours of the likelihood test statistics of 2EG provided by Thompson et al. (1996). It is clear that the only source within the 95% confidence contour (which is ussually considered as representative of the EGRET statistical point source uncertainties) is PKS 1830-211. The angular separation between the location of the highest likelihood test statistic for 2EG and the position of the AGN is just arcminutes. The remaining radio sources are too far from the EGRET source position to significantly contribute to the -ray emission. The nearest source is , and one might speculate that the pulsar produced in the supernova event that originated the remnant could be a Geminga-like object with a high proper velocity that would have driven it to the position of the EGRET detections. This seems unlikely. In fact, the angular distance from the flux peak of to the EGRET highest confidence contour is , which corresponds to a distance of pc. The proper motion of the pulsar then should exceed 1300 km s-1, which seems unrealistic. Moreover, the pulsar hypothesis is not supported by the observed -spectrum of 2EG (Merck et al. 1996). This spectrum can be properly fitted by a power-law given by cm-2 s-1 MeV-1, which is considerably steeper than observed in -ray pulsars (indices typically , see Fierro et al. 1993). The possible presence of -ray variability in the time history of 2EG (see below) also suggest to rule out a pulsar origin of the emission.
It is also worhty to remark that the Bayes's theorem has been used by Mattox et al. (1997) to computed a posteriori probability that PKS 1830-211 is the correct identification of 2EG . The resulting probability is 0.98, with an a priori probability of . In this computation radio sources from the Parkes-MIT-NRAO survey (Griffith & Wright 1993) were used. If we take into account that there are just 183 strong flat-spectrum radio sources at the possibility of a chance association results negligible.