Quasars are among the most energetic objects in the Universe. We now know that they live at the centers of galaxies and that they are the most dramatic manifestation of the more general phenomenon of active galactic nuclei (AGNs). These include a wide variety of exotica such as Seyfert galaxies, radio galaxies, and BL Lacertae objects. Since the discovery of quasars in 1963, much effort has gone into understanding their energy source. The suite of proposed ideas has ranged from the relatively prosaic, such as bursts of star formation that make multiple supernova explosions, to the decidedly more colorful, such as supermassive stars, giant pulsars or ``spinars,'' and supermassive black holes (hereinafter BHs). Over time, BHs have gained the widest acceptance. The key observations that led to this consensus are as follows.
Quasars have prodigious luminosities. Not uncommonly, L ~ 1046 erg s-1; this is 10 times the luminosity of the brightest galaxies. Yet they are tiny, because they vary on timescales of hours. From the beginning, the need for an extremely compact and efficient engine could hardly have been more apparent. Gravity was implicated, because collapse to a black hole is the most efficient energy source known. The most cogent argument is due to Donald Lynden-Bell (1969, Nature, 223, 690). He showed that any attempt to power quasars by nuclear reactions alone is implausible. First, a lower limit to the total energy output of a quasar is the energy, ~ 1061 erg, that is stored in its radio-emitting plasma halo. This energy weighs 1040 g or 107 M. But nuclear reactions produce energy with an efficiency of only = 0.7%. Then the waste mass left behind in powering quasars would be at least MBH 109 M. Lynden-Bell argued further that quasar engines are 2R 1015 cm in diameter because large variations in quasar luminosities are observed on timescales as short as 10 h. But the gravitational potential energy of 109 M compressed into a volume as small as 10 light hours is ~ G MBH2 / R 1062 erg. As Lynden-Bell noted, ``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.'' We now know that the total energy output is larger than the energy that is stored in a quasar's radio source; this strengthens the argument. Meanwhile, a caveat has appeared: the objects that vary most rapidly are now thought to contain relativistic jets that are beamed at us. This boosts the power of a possibly small part of the quasar engine and weakens the argument that the object cannot vary on timescales less than the light travel time across it. But this phenomenon would not occur at all if relativistic motions were not involved, so BH-like potential wells are still implicated. These considerations suggest that quasar power derives from gravity.
The presence of deep gravitational potentials has long been inferred from the large velocity widths of the emission lines seen in optical and ultraviolet spectra of AGNs. These are typically 2000 to 10,000 km s-1. If the large Doppler shifts arise from gravitationally bound gas, then the binding objects are both massive and compact. The obstacle to secure interpretation has always been the realization that gas is easy to push around: explosions and ejection of gas are common astrophysical phenomena. The observation that unambiguously points to relativistically deep gravitational potential wells is the detection of radio jets with plasma knots that are seen to move faster than the speed of light, c. Apparent expansion rates of 1 - 10 c are easily achieved if the true expansion rate approaches c and the jet is pointed almost at us.
The final pillar on which the BH paradigm is based is the observation that many AGN jets are well collimated and straight. Evidently AGN engines can remember ejection directions with precision for up to 107 yr. The natural explanation is a single rotating body that acts as a stable gyroscope. Alternative AGN engines that are made of many bodies - like stars and supernovae - do not easily make straight jets.
A variety of other evidence also is consistent with the BH picture, but the above arguments were the ones that persuaded a majority of the astronomical community to take the extreme step of adopting BHs as the probable engine for AGN activity. In the meantime, BH alternatives such as single supermassive stars and spinars were shown to be dynamically unstable and hence short-lived. Even if such objects can form, they are believed to collapse to BHs.
The above picture became paradigm long before there was direct evidence for BHs. Dynamical evidence is the subject of the present and following articles. Meanwhile, there are new kinds of observations that point directly to BH engines. In particular, recent observations by the Advanced Satellite for Cosmology and Astrophysics (ASCA) have provided strong evidence for relativistic motions in AGNs. The X-ray spectra of many Seyfert galaxy nuclei contain iron K emission lines (rest energies of 6.4 - 6.9 keV; see Figure 1). These lines show enormous Doppler broadening - in some cases approaching 100,000 km s-1 or 0.3c - as well as asymmetric line profiles that are consistent with relativistic boosting and dimming in the approaching and receding parts, respectively, of BH accretion disks as small as a few Schwarzschild radii.
The foregoing discussion applies to the most powerful members of the AGN family, namely quasars and high-luminosity Seyfert and radio galaxies. It is less compelling for the more abundant low-luminous objects, where energy requirements are less demanding and where long jets or superluminal motions are seen less frequently or less clearly. Therefore a small but vocal competing school of thought continues to argue that stellar processes alone, particularly those that occur during bursts of star formation, can reproduce many AGN characteristics. Nonetheless, dynamical evidence suggests that BHs do lurk in some mildly active nuclei, and, as discussed in the next article, even in the majority of inactive galaxies.
Figure 1. A composite x-ray spectrum of Seyfert nuclei taken with ASCA showing the relativistically broadened Fe K line. The solid line is a fit to the line profile using two Gaussians, a narrow component centered at 6.4 keV and a much broader, redshifted component. [Figure adapted from Nandra, K., et al. Astrophys. J. 477, 602 (1997).]