The Catalogs

The Deep X-ray Radio Blazar Survey (DXRBS) uses a cross-correlation of all serendipitous sources in the publicly available database of ROSAT sources, WGACAT (White, Giommi & Angelini 1995), having quality flag (to avoid problematic detections) with a number of publicly available radio catalogs. North of the celestial equator, we used the 20 cm and 6 cm Green Bank survey catalogs NORTH20CM and GB6 (White & Becker 1992; Gregory et al. 1996), while south of the equator, we used the Parkes-MIT-NRAO catalog PMN (Griffith & Wright 1993). All sources with radio spectral index at a few GHz were selected as blazar candidates.

For objects north of the celestial equator, 6-20 cm radio spectral indices were obtained directly from the cross-correlation of the GB6 and NORTH20CM catalogs. For sources at southern declinations, the lack of a comparably deep radio survey at a second frequency required a different strategy. In the band , we cross-correlated the sources with the public NVSS database (Condon et al. 1997); our selection of candidates is still not completed in this declination range, since the NVSS is not yet 100% complete. Further south, the positional accuracy of the NVSS, which covers the sky north of , decreases somewhat (Condon et al. 1997). In this region, we conducted a snapshot survey with the Australia Telescope Compact Array (ATCA) at 3.6 and 6 cm (note that the positional accuracy of the ATCA snapshots deteriorate significantly at due to the East-West nature of the array), to get also radio spectral indices unaffected by variability (see § 7). The first set of ATCA observations (of 163 X-ray/radio sources) took place 11-13 November 1995. A second set of 55 X-ray/radio sources (some of which have preliminarily been classified as blazar candidates based upon the 6-20 cm spectral index computed from their PMN and NVSS fluxes) were observed in October 1997 to complete the coverage of the southern sample. These ATCA observations will be discussed in a future paper. We had originally requested observations at 6 and 20 cm at the ATCA as well, to match our northern sample, but the time allocation committee decided otherwise, based on the instrumental configuration. Note that, as the NVSS has a much smaller beam size than the GB6 survey, it is preferable to use, whenever possible, the NORTH20CM 20 cm fluxes to derive spectral indices. Extra care was taken in the region, where we had to resort to the NVSS for this purpose, to include the flux from all sources in a 2 arcmin radius. We stress, however, that this problem is severe only for extended, steep-spectrum radio sources, and not for the core-dominated, flat-spectrum sources we are interested in.