Black holes (BHs) progressed from a theoretical concept to a necessary ingredient in extragalactic astronomy with the discovery of quasars by Schmidt (1963). Radio astronomy was a growth industry at the time; many radio sources were identified with well-known phenomena such as supernova explosions. But a few were identified only with "stars" whose optical spectra showed nothing more than broad emission lines at unfamiliar wavelengths. Schmidt discovered that one of these "quasi-stellar radio sources" or "quasars", 3C 273, had a redshift of 16% of the speed of light. This was astonishing: the Hubble law of the expansion of the Universe implied that 3C 273 was one of the most distant objects known. But it was not faint. This meant that 3C 273 had to be enormously luminous - more luminous than any galaxy. Larger quasar redshifts soon followed. Explaining their energy output became the first strong argument for gravity power (Zel'dovich 1964; Salpeter 1964).
Studies of radio jets sharpened the argument. Many quasars and lower-power active galactic nuclei (AGNs) emit jets of elementary particles that are prominent in the radio and sometimes visible at optical wavelengths. Many are bisymmetric and feed lobes of emission at their ends (e.g., Fig. 1). Based on these, Lynden-Bell (1969, 1978) provided a convincing argument for gravity power. Suppose that we try to explain the typical quasar using nuclear fusion reactions, the most efficient power source that was commonly studied at the time. The total energy output of a quasar is at least the energy stored in its radio halo, E ~ 1054 J. Via E = mc2, this energy weighs 107 solar masses (M). But nuclear reactions have an efficiency of only 0.7%. So the mass that was processed by the quasar in order to convert 107 M into energy must have been 109 M. This waste mass became part of the quasar engine. Meanwhile, rapid brightness variations showed that quasars are tiny, with diameters 2R 1013 m. But the gravitational potential energy of 109 M compressed inside 1013 m is GM2 / R ~ 1055 J. "Evidently, although our aim was to produce a model based on nuclear fuel, we have ended up with a model which has produced more than enough energy by gravitational contraction. The nuclear fuel has ended as an irrelevance" (Lynden-Bell 1978). This argument convinced many people that BHs are the most plausible quasar engines.
Figure 1. Cygnus A at 6 cm wavelength (Perley, Dreher, & Cowan 1984). The central point source is the galaxy nucleus; it feeds oppositely directed jets (only one of which is easily visible at the present contrast) and lobes of radio-emitting plasma. The resolution of this image is about 0."4.
Jets also provide more qualitative arguments. Many are straight over ~ 106 pc in length. This argues against the most plausible alternative explanation for AGNs, namely bursts of supernova explosions. The fact that jet engines remember ejection directions for 106 yr is suggestive of gyroscopes such as rotating BHs. Finally, in many AGNs, jet knots are observed to move away from the center of the galaxy at apparent velocities of several times the speed of light, c. These can be understood if the jets are pointed almost at us and if the true velocities are almost as large as c (Blandford, McKee, & Rees 1977). Observations of superluminal motions provide the cleanest argument for relativistically deep potential wells.
By the early 1980s, this evidence had resulted in a well-established paradigm in which AGNs are powered by BHs accreting gas and stars (Rees 1984). Wound-up magnetic fields are thought to eject particles in jets along the rotation poles. Energy arguments imply masses M ~ 106 to 109.5 M, so we refer to these as supermassive BHs to distinguish them from ordinary-mass (several-M) BHs produced by the deaths of high-mass stars. But despite the popularity of the paradigm, there was no direct dynamical evidence for supermassive BHs. The black hole search therefore became a very hot subject. It was also dangerous, because it is easy to believe that we have proved what we expect to find. Standards of proof had to be very high.