Lensing remained an interesting theoretical possibility for most of the twentieth century. The deflection of light by a mass was calculated classically by Soldner (1801) and correctly by Einstein as a consequence of general relativity, and famously observed by Eddington in 1919 using position shifts of stars close to the Sun during a solar eclipse. The possibility of multiple images formed by individual objects was considered by Chwolson (1924). However, it was not until 1979 that the first gravitational lens system was actually found.
The first lens system, Q0957+561, was discovered by Walsh et al. (1979) during the course of optical followup of sources found in a radio survey at 966 MHz with the Jodrell Bank telescope. The source object in this system is a radio-loud quasar at redshift 1.41, doubly imaged by a galaxy at redshift 0.36 (Fig. 1). The large (6") separation of the two images in this system is due to the fact that the lensing galaxy is assisted by a cluster at the same redshift; this allowed Walsh et al. to measure separately the redshifts of the two lensed images, whose spectral similarity supported the hypothesis that their light originates from the same background object. In the next few years, further objects were discovered, many of them in radio surveys; these have the advantage that the sources they contain are predominantly non-thermal emitters such as quasars, without contamination from stellar processes. A typical non-thermal radio source consists of a central bright flat-spectrum core, corresponding to the active nucleus at the centre of the host galaxy, and extended steep-spectrum emission in lobes which result from ejection of material from the active centre. Although the Q0957+561 system has a double image of the core, further quasar lenses were found in which the extended radio emission was gravitationally imaged, resulting in rings (e.g. MG1131+0456, Hewitt et al. 1988).
Figure 1. A modern view of the Q0957+561 lens system, made with the e-MERLIN telescope at 5GHz. The quasar consists of a core, radio jet (top left), and lobe (bottom right). The second image of the core (bottom of the image) is close to the lensing galaxy. (Image credit: Muxlow, Beswick & Richards, Manchester).
Searches for lensed radio sources divided into two main areas. The first, following the success of the MG survey, targeted extended radio lobes, looking for rings associated with the presence of galaxies in front of them (Lehár et al. 2001, Haarsma et al. 2005). The second method involved systematic targeting of flat-spectrum radio sources, in which the central bright point component dominates the radio emission. This makes lensing relatively easy to recognise, although the low optical depth to lensing means that a large number of candidates must be examined to find relatively few lenses. The CLASS survey (Myers et al. 2003, Browne et al. 2003) is still the largest systematic radio survey, and produced 22 lens systems from a parent sample of 16503 northern objects initially observed with the VLA and followed up at higher resolution using Merlin and VLBI. A southern extension of this survey also exists (e.g. Winn et al. 2001) which discovered a further 4 lens systems. Most of the lenses are elliptical galaxies, since these are generally more massive and hence dominate the lensing cross-section, but a significant minority are later types. Subsamples of these 22 are still important for many of the astrophysical and cosmological applications that will be discussed later in this review.
About 90% of quasars do not have bright radio emission. Searches for lensed radio-quiet quasars began soon after the Walsh et al. (1979) initial discovery, most successfully using the HST. Lensed optical quasars began to be discovered in substantial numbers following the availability of the Sloan Digital Sky Survey (SDSS, York et al. 2000) and subsequently of quasar catalogues derived from it. The Sloan Quasar Lens Search (SQLS; Oguri et al. 2006, Inada et al. 2008, Inada et al. 2010, Inada et al. 2012) has discovered the majority of these, by following up quasars which show extended emission on SDSS images, suggesting the presence of secondary lensed images, or of a lensing galaxy, or both. The SQLS has produced about 30 quasar lenses in the most recent catalogue, including some of the widest-separation lenses known (Inada et al. 2003, Inada et al. 2006). In addition a number of smaller surveys have increased the number of lenses, using various methods. These include the use of higher-image quality supplementary surveys such as UKIDSS (Lawrence et al. 2007) to find small-separation or high-flux-ratio lenses (Jackson et al. 2009, 2012).
The most complete current database, the Masterlens project (Moustakas et al. 2012), lists 120 quasar lenses of which 71 result from two surveys (SQLS and CLASS) and the remainder are serendipitous discoveries or result from smaller surveys. It is this sample that forms the current basis of the scientific results discussed in this review. Future instruments will increase this sample by many orders of magnitude, allowing correspondingly more detailed and wide-ranging science which is discussed in the final section.