The preliminary VLA FIRST catalog of radio sources covered 306 in a narrow strip through the north Galactic pole (BWH) and contained sources, complete to a flux density limit of 1 mJy for point sources. Comparisons between the FIRST catalog and standard radio calibration sources indicate that the systematic astrometry errors are in both RA and Dec. Extensive tests also indicate that even for the faintest radio sources, positions are accurate to (90% confidence). Nonetheless, in matching to optical counterparts we required only 2" agreement in position, in part to allow an independent check on the radio positional accuracy (see section IV).
QSO candidates were selected by matching FIRST survey sources to the APM catalog of the Palomar Observatory Sky Survey (POSS I). Objects classified by the APM as stellar on either of the two POSS I emulsions, O (blue) or E (red), brighter than 17.5 magnitude on the E plate, and within 2" of a FIRST radio source were included in this pilot quasar survey. In keeping with the desire to avoid optical selection effects, no color cut was imposed on this initial candidate list.
The original National Geographic-Palomar Observatory Sky Survey was carried out using Eastman 103a-O and 103a-E emulsions. Plots of the effective system response for each are given in Minkowski & Abell (1963). The ``O'' and ``E'' passbands have effective wavelengths of roughly 4200Å and 6400Å, and effective widths of approximately 1200Å and 400Å, respectively. Except for the narrowness of the E passband, these are similar to Johnson B and Cousins R; the color transformation for normal stars is (B-R) = 0.875(O-E) + 0.073 (Bessell 1996, private communication). In what follows, we have chosen not to transform the magnitudes from O and E to a more standard system for two reasons. Objects with emission lines will not transform in a straightforward manner as do stars. Also, the errors in O and E are rather large. In a comparison of 33 QSOs for which the Automated Plate Scanner (http://isis.spa.umn.edu) magnitudes were available on-line, we obtain a scatter of 0.4 magnitudes in both O and E and 0.35 magnitudes in O-E. Much of the scatter can perhaps be attributed to the lack of field-by-field calibration of the APM catalog. Within this error range, O and E can be considered approximately equal to B and R magnitudes over the color range of our sample. We have begun an observing program at Lick Observatory to improve the optical photometry of the QSOs in the sample, however, the completeness of the survey remains problematic because of the large errors. We will address this issue in more detail in future installments of this survey.
The other selection criteria also raise some issues regarding completeness. The requirement of a 2" positional coincidence discriminates against QSOs associated with extended radio sources (lobes), selecting only those QSOs with nuclear radio emission or compact morphology as seen by the FIRST survey. Additional QSOs might be found by relaxing the coincidence requirements but only at the expense of additional chance coincidences. Part of the intent of the pilot survey is to explore the level of completeness arising from the morphological considerations.
With the above criteria, 219 QSO candidates were selected, 0.8% of the total FIRST catalog. Ninety-seven were classified as stellar on both POSS I emulsions. Another 29 were classified as stellar on the E plates only and 93 as stellar on the O plates only. Since the O magnitudes are often much fainter than the E magnitudes, their classification is typically less reliable, especially near the plate limit where the tendency is for the APM to classify these objects as stellar. This leads to a large number of apparently normal, red galaxies being included in the candidate list.
Using the NASA Extragalactic Database (NED), 25 of the 219 candidates were found to be already identified in the literature as QSOs or galaxies, leaving 194 objects for spectroscopic follow-up.