The optical identification of the first discrete extraterrestrial radio source occurred as a result of a telephone call from J. S. Hey to the Royal Greenwich Observatory on the afternoon of February 28, 1942. Recognizing that the source of extensive jamming of British radar over the previous two days appeared to follow the Sun, Hey was delighted to learn that an unusually large sunspot had just transited the solar disk; despite its skeptical reception by his superiors, Hey's identification proved correct (Hey 1973).
In the ensuing decade, progress in the detection of new extrasolar radio emitters far outstripped the ability of astronomers to associate them with optical counterparts. The first breakthrough came in 1949 when Bolton, Stanley and Slee (1949) identified the Crab Nebula, M87, and NGC 5128 (Cen A) with three of the brightest radio sources in the sky, although they concluded that the bizarre morphology of the latter generally favored a Galactic interpretation for radio emitters since ``the probability of [such] an unusual object in our own Galaxy seems greater than a large accumulation of such objects at a great distance.'' The following year, Ryle, Smith, and Elsmore (1950) concurred in this conclusion despite finding 0/146 bright (4.0) stars, 0/21 novae, 0/38 planetary nebulae, 0/29 diffuse Galactic nebulae, and 4/5 of the brightest galaxies coincident with entries in their fifty-source radio catalog. It was not until the classic papers of Baade & Minkowski (1954a,b), which among other things pronounced Cygnus A ``an extragalactic affair'', that the era of extragalactic radio source identification can be said to have begun.
The largest radio catalogs in existence prior to 1995 contained, in total, approximately 100,000 distinct entries. In striking contrast to the earliest speculations, fewer than 20 of these relatively bright radio sources are identified as stars. Indeed, fewer than 1000 stellar radio detections have been made despite decades of sensitive, targeted searches (Hjellming 1988 and references therein; Wendker 1995), and of all cataloged radio sources are Galactic objects. A search of the NED database, however, suggests that the fraction of identified extragalactic radio emitters today is little better than it was in 1950, when 7 out of 67 known radio sources had identified counterparts (Baade and Minkowski 1954b). The problem now is the same as it was 50 years ago: the angular resolution of large-area radio surveys is generally too poor (several arcminutes) to allow for the unambiguous association of cataloged objects with individual optical counterparts which, at the limit of the POSS-I plates, number 4000 deg at high Galactic latitudes. Interferometric surveys at centimeter wavelengths can achieve the desired positional accuracy of but have, until recently, covered only deg of sky, resulting in fewer than 500 optical identifications for radio-selected objects at faint flux levels (see Table 5).
While considerable information on the class and emission mechanism of a radio source can be derived from observations of its radio morphology, its spectrum, and its polarization characteristics, optical observations are still required to establish the source's distance and to classify it unambiguously. Eight years ago, we began to construct Faint Images of the Radio Sky at Twenty-cm (FIRST) with the primary goal of obtaining a very large sample of radio sources with positions sufficiently accurate that the majority of objects detected could be easily identified on the basis of positional coincidence alone. FIRST has been designed to cover the same 10,000 deg region as the Sloan Digital Sky Survey (SDSS), which will obtain deep optical images of the northern sky in five colors, and will take spectra of a million objects over the next decade (Gunn and Knapp 1993). However, an archive of the optical sky in the form of the National Geographic-Palomar Observatory Sky Survey (POSS) plates already exists. The identification of even 20% of all FIRST sources with counterparts at or above the POSS plate limit will immediately provide samples of various radio source populations from one to three orders of magnitude larger than those in existence, and will advance significantly our knowledge of the radio universe.
In this paper, we present the results of an optical identification program for FIRST radio sources based on the Cambridge Automated Plate Measuring Machine (APM) scans of the POSS I plates (McMahon and Irwin 1992). In § 2, we describe briefly the parameters of the FIRST survey and the catalog derived therefrom. In § 3, we describe the genesis of the APM catalog and discuss its astrometric, photometric, and source classification uncertainties with particular attention to those which either 1) the FIRST results can help to refine, or 2) are particularly relevant to radio source identification; the following section (§ 4) demonstrates the utility of FIRST as an astrometric calibrator by deriving both intraplate and plate-to-plate corrections to the optical astrometric solution. Section 5 describes the matching algorithms employed and discusses such issues as false match rates, reliability and completeness of the proposed identifications, and the effects of optical and radio source morphology. Section 6 discusses color, magnitude, and radio flux density distributions for the more than 70,000 counterparts we have identified, while the following section (§ 7) describes the format of the counterpart catalog (the full contents of which are available on the Web). In the discussion (§ 8), we place in context this identification program for faint radio sources, and summarize the statistics of the populations represented. A reprise of our results and a precis of future work concludes our report (§ 9).