Annu. Rev. Astron. Astrophys. 1992. 30: 311-358 Copyright © 1992 by Annual Reviews. All rights reserved

7. SEARCHES FOR GRAVITATIONAL LENSES

The first examples of multiple quasars, rings, and arcs were identified in observing programs directed toward quite different goals. Since then, roughly half of the known instances of gravitational lensing have arisen serendipitously, which makes it difficult to quantify the incidence of lensing and to draw cosmographic deductions. In this section we briefly review attempts to search for lenses in uniform, complete samples of extragalactic sources.

7.1 Radio Surveys

A significant number of lensed objects have been found from radio surveys. The MIT-Greenbank (MG) survey (Bennett et al. 1986) consists of 6000 sources found at 5 GHz using a transit telescope. These were subsequently mapped at the VLA with 0.3" resolution and those that exhibited unusual compact structure were imaged optically. Subsequent spectroscopy of the nonstellar objects has yielded two secure cases of multiple imaging (plus four candidates) and three rings (Burke 1990).

Gravitational lenses have also been sought in sources observed with MERLIN (Walsh 1992) which is now an ideal instrument for finding candidates with subarcsecond structure. One ring has been found. Radio ring images of background extended radio sources ought to be more common in existing surveys than has hitherto been reported (Kochanek & Lawrence 1990) and there is optimism that many more of these sources, which are promising for measuring the Hubble constant (cf Section 4.1), will be found.

Q0957+561 and 1830-211 were also found from radio surveys.

7.2 Optical Surveys

Existing quasar surveys (e.g. Boyle et al. 1988) show that the number of quasars brighter than some luminosity L varies as L^-, where ~ 2.7 (Boyle et al. 1988) down to ~ 19^m. As the dominant lens cross sections (due to fold caustics, cf Section 3.3) decrease only as the inverse square of the magnification, the majority of lensed sourcess brighter than 19^m ought to be highly magnified as a result of magnification bias (Turner et al. 1984, Section 6.5). This has so far proved to be the case, and has motivated optical lens searches among samples of bright quasars.

Several groups (e.g. Djorgovski & Meylan 1989, Surdej 1990, Swings et al. 1990) have combined to search for lenses in an ESO key project. CCD images of approximately 400 bright (M_V -29), distant (z 1) quasars drawn from highly heterogeneous and ill-defined catalogs have been obtained. Two secure and two good candidate lenses have so far been identified, as well as two quasar pairs (cf Section 2.1). Negative searches in smaller bright quasar samples have also been reported (Crampton 1992, Yee 1992). Selection criteria are somewhat poorly defined, which leads to difficulties in drawing quantitative conclusions about the frequency of multiple imaging (cf Section 5.2.2).

These shortcomings may be partly overcome in the APM survey (Webster et al. 1988b, Hewett 1992) which establishes a methodology for creating a complete sample. Quasar candidates are obtained by uniformly scanning photographic plates covering a large fraction of the southern sky. This survey appears to be highly efficient at rediscovering known quasars (Foltz et al. 1989) and 2500 quasars brighter than 20^m are believed to be present in the survey. But the survey may be less efficient at identifying multiple quasars with separations similar to the angular resolution because, by selecting unusually elliptical images, it is prejudiced against small separations and, in particular, five image configurations (Kochanek 1991b). One lens candidate with angular separation 5" [Q1429-008 (Hewett et al. 1989)], and three quasar pairs have so far been found using this method. Further success is dependent upon obtaining spectra of a large number of candidate objects.

A comparatively new search technique uses the Hubble Space Telescope to make short exposures during gaps in the regular observing program of high luminosity quasars (Bahcall et al. 1992). 354 bright quasars will be surveyed. Simulations show that, despite the degraded point spread function, this search ought to be sensitive to separations as small as ~ 0.1" and magnitude differences up to ~ 2^m. As of this writing, one lens candidate has been found [Q1208+101 (Maoz et al. 1992)].