2.2. Optical Emission Lines, Variability, and Zero Proper Motion Selection

There have been a wide variety of AGN surveys relying on low-resolution optical spectra. These techniques select AGN on the strength of the UV/optical emission lines (see Osmer & Hewett 1991 for a summary of this technique, and Salzer et al. 2000 for the latest application with the Kitt Peak National Observatory International Spectroscopic Survey). While this method essentially has no "false" detections, its completeness is very difficult to evaluate, since the signal-to-noise depends on the equivalent widths of the lines. Since the lines observed in the optical depend on the redshifts of the objects (e.g., from H to Ly) and vary in intrinsic strength by over an order of magnitude, it is very difficult to evaluate the completeness of the sample. Furthermore, the effects of redshifting make the effective band smaller by (1 + z), such that at z = 3, only 1400 Å in the rest-frame are covered by the total ground-based optical wavelength band (3200 - 8900 Å). The limit of this technique (Ho, Filippenko, & Sargent 1995) is to obtain high signal-to-noise spectra of the nucleus of every object in a complete sample. However, even this method runs into the difficulty of removing absorption lines due to stars in the relatively low-luminosity AGN spectra and requires extreme care.

Variability can also be used as a survey technique (Usher & Mitchell 1978; Brunzendorf & Meusinger 2002; Dobrzycki et al. 2003), since on long enough timescales, virtually all AGN are variable (Veron & Hawkins 1995; Giveon et al. 1999). This method requires a large number of observations and seems to produce a lower areal density of objects than other techniques. However, this method also finds large numbers of variable stars and supernovae and must be combined with other selection criteria.

Similarly, since AGN are at great distances compared to bright stars, the absence of proper motion can be a guide to AGN selection. However, this technique has not been widely used and fails at faint magnitudes, where most of the stars also do not show proper motion. This technique will be resurrected when large solid angle astrometric data sets are available.