(This essay was originally written as a commission for Astronomy magazine. However, by the time I had included the topics of current interest and done them minimal justice, it came out four times longer than they wanted, so most of this text ended up cut or rewritten. There are lots of interesting things that didn't make the printed version, so here is the Director's Cut.)

Forty years ago, the unexpected discovery of quasars showed astronomers just how surprising the Universe could be, and set us on new journeys of exploration in directions few could have foreseen. Finding the first examples required a partnership between radio astronomy and more traditional research with the world's (then) largest optical telescope, on Palomar Mountain. Credit for the final leap identifying these objects as being at high redshifts, and therefore the most powerful objects we know in the Universe, goes to Maarten Schmidt, recorded in an appropriately "cosmic" portrait on the cover of Time magazine for March 11, 1966. Efforts to understand these powerhouses included many idea that sounded like sheer science fiction, and all of them were at the edge of our understanding. Indeed, the very word "quasar" now carries a whiff of the mysterious and exotic. It can be found as a brand, company, or project name in such disparate fields as television sets, software development, publishing, consulting, diamond trading, Internet service providers, and the bicycle my older son recently won as a door prize. Astronomers have been studying quasars and related phenomena intently for forty years now. Has our understanding reached the level of maturity that we associate with turning forty?

It became clear during the 1960s that quasars fall within a context including other, related kinds of objects. These are all collectively known as "active galactic nuclei" or AGN, and are linked by showing powerful energy release from a small area in the center of a galaxy, well beyond what ordinary stars and their lifecycles (including supernovae and neutron stars) can account for. Among these, the defining properties of quasars are a large redshift (implying great distance and high luminosity), and that they appear starlike in ordinary telescopic photographs (so that the nucleus outshines any surrounding galaxy). The original derivation of "quasar" was from the acronym QSRS for quasi-stellar radio source, since the first few known were all found as a result of their radio emission. However, similar objects with very weak (or undetectable) radio emission outnumber the radio-loud quasars; all these together are known as quasistellar objects or QSOs, but "quasar" is often employed to mean the whole bunch of them when the distinction about radio emission is not relevant. The spectra of quasars are quite different from those of ordinary galaxies, showing broad emission lines of gas excited to high levels, and an underlying blue continuous spectrum lacking the absorption lines from ordinary stars. A similar pattern is seen, at lower power levels, from the nuclei of Seyfert galaxies, which contain tiny active nuclei that are less brilliant than quasars but otherwise share many of their properties. There are also galaxies which share the jets and twin lobes of radio-emitting material seen in radio-loud quasars; some of these show similar spectral lines, but some show only normal starlight, so the action is either invisible or hidden to optical light. In one subset of AGN, the BL Lacertae objects or blazars, the spectrum shows a featureless continuum, well understood if we are looking right down the jets. The beaming of radiation from material moving close to the speed of light means, in these objects, that the jet's light is boosted to overwhelm everything else, even the powerful emission lines from the nucleus.

Some of the family resemblances among active galactic nuclei have led to a standard consensus picture for their energy generation, one which admittedly owes its popularity partly to the fact that no other scheme can explain what we see any better, although the direct evidence in its favor remains limited. The variability in light output (and, for that matter, radiation from gamma rays to radio wavelengths) shows that most of the radiation in active nuclei must be coming from tiny regions, no more than light-hours in extent. Otherwise, we couldn't see the objects vary as rapidly as we do, with our view of the whole object smeared by the time it takes light to cross it. A strong gravitational field would help explain how the spectral lines could be so greatly broadened by Doppler shifts, often indicating gas moving around at velocities in excess of 5000 km/s, without having long ago left the nucleus completely. High temperatures, or other ways of imparting large amounts of energy to the emitting material, are needed to account for the strong X-rays seen from all kinds of AGN. Finally, the extent of neatly collimated radio jets, some of which stay on the straight and narrow for a couple of million light-years out from the nuclei, shows that the central source has both preferred directions to throw material out (and do so near the speed of light), and a memory spanning millions of years as to which directions these are.

Among the objects known or reasonably theorized about in astrophysics, the best explanation of all these facts seems to lie in an enormous black hole and its surroundings - a black hole incorporating not the paltry few solar masses from a collapsed star, but millions or billions of solar masses, grown in the dense core of a galaxy. Gas close to such a monster would follow the same pattern as particles in a planetary ring, gas and dust around a young star, or gas in a forming galaxy, flattening into an accretion disk orbiting the black hole. Collisions among atoms in this disk would naturally heat them to very high temperatures, especially close to the black hole where orbital velocities rise toward the speed of light. As a bonus, several sets of calculations have shown that even small interstellar magnetic field could be amplified in such a disk, so that they would launch some of the matter in twin jets perpendicular to the disk, at nearly the speed of light. This general picture of active nuclei - hot accretion disks around supermassive black holes - is widely accepted, but we have some trouble connecting specific pieces of it to phenomena we can actually see. In a general way, we expect the X-rays to come from the hot inner parts of the disk, and ultraviolet and visible continuum radiation to come from farther out where the disk is cooler. The strong emission lines must come from clouds of gas lit up by the ultraviolet and X-ray emission from closer in. The radio jets have an obvious connection; in some cases, radio interferometry has traced them to within a light-year of the core, not much bigger than the scale of the accretion disk. As we will see, though, while dynamical signs of the massive black holes are abundant, direct evidence of the accretion disk remains elusive.

As in fields from political polling to wildlife biology, we need to know what complete samples of quasars are like before we can claim to understand this population. The first quasar discoveries, were, understandably, rather hit-or-miss in nature. Some were found as "stars" identified with strong radio sources, some as faint blue stars, and a few were catalogued as irregularly variably stars before their spectra revealed their true nature (so V396 Herculis isn't a star at all, even if it is variable). Even as techniques matured to select quasars for having particularly blue colors (since faint blue stars are uncommon), or identify larger samples of radio sources, various approaches were applied to various pieces of the sky, so that a truly complete census was still lacking.

New techniques have made it possible to find large samples of astronomical objects (including active galactic nuclei) with statistically well-defined selection processes. These are, above all, based on surveys of large areas of the sky at various wavelengths, and the ability to winnow desired kinds of objects based on color or image structure from a huge collection of detected objects. For active nuclei, there have been important contributions from deep radio surveys (especially the two carried out concurrently with the Very Large Array, known as the FIRST and NVSS projects), the all-sky X-ray survey conducted as the primary mission of the ROSAT satellite, the all-sky mapping in the near-infrared by the 2MASS project, and the ongoing Sloan Digital Sky Survey in the optical.

Impatient with waiting for some of the survey groups to do so, New Zealand amateur astronomer Eric Flesch put together an extensive catalog of over 100,000 possible QSOs based on associations among catalogued X-ray and radio sources, which matched starlike objects from the US Naval Observatory scans of the Palomar Sky Survey. Listed on his FTP site as Free-Lunch, this has proven statistically useful in its own right.

The Sloan and 2MASS samples of QSOs rely on the fact that, especially as one examines ever wider wavelength spans, the colors of QSOs at various redshifts don't match those of any kinds of ordinary stars. Using what we know about the colors of recognized active nuclei, and various kinds of stars, the survey teams can automatically select not just objects likely to be quasars, but objects likely to be quasars at particular redshifts (such as the highest redshifts, especially interesting for a variety of reasons). The infrared surveys can pick up any population of quasars which is much redder than we have come to expect, whether because of surrounding dust or intrinsically different radiation properties. It has long been recognized that earlier selection techniques, based either on blue colors or the emission-line spectrum in the optical region, might be biased against finding such red objects, so this could be an important new window.

Indeed, there is a substantial population of red quasars. Roc Cutri at Caltech's Infrared Processing and Analysis Center has led a group using the 2MASS survey data to find such objects, turning up large numbers of red quasar candidates and thereafter confirming many of them spectroscopically. These objects turn out to be about as numerous as the ordinary quasars with traditional bluer colors. Their differences can be explained if this redder population consists of quasars where the nucleus is partly obscured by surrounding dust (and gas, seen as it absorbs some of their X-rays). In fact, much of the quasar light that we do see from these particular nuclei shows traces of being scattered off this dust, rather than seen directly - in giant reflection nebulae that allow us a view of regions too deeply shrouded for us to see directly. It turns out, from the Chandra X-ray observations of some of these red quasars, that they may be the missing link in connecting active nuclei to the total X-ray brightness of the sky at high energies, which not only solves a long-standing puzzle in X-ray astronomy but fills the gap in understanding the history of active galaxies.

000703.6+155423	004118.7+281640	024807.3+145957	030558.6-115315	082311.3+435318
z=0.114	z=0.194	z=0.072	z=0.333	z=0.182

2MASS three-color composite images of Red AGN (Cutri R.M., Nelson, B.O., Kirkpatrick, J.D., Huchra, J.P. and Smith, P.S. 2001, in "The New Era of Wide-Field Astronomy," ASP Conference Series, Vol 232, 78, R. Clowes, A. Adamson and G. Bromage eds.). All images cover 2'x2', north is to the top, east is to the left. Blue = 2MASS J-band, Green = 2MASS H-band, Red = 2MASS Ks-band.

Clues to the nature of active nuclei have long been sought from their environments, especially the surrounding host galaxies. If a direct assault on their physics proved difficult, perhaps their favorite haunts could yield some of their secrets. From their definition, it is clear that surveying galaxies around quasars will be difficult, since they show no such galaxies on ordinary pictures. Thus, for decades, looking at the host galaxies of AGN meant Seyfert and radio galaxies. These offered rich pickings, with many objects available and some quite nearby and furnishing us with detailed views. Seyfert nuclei occur mostly in spirals having a large, bright central bulge, galaxies informally known as "early" in Hubble type. A suspiciously large number of Seyfert nuclei lie within galaxies that are strongly disturbed or closely paired with another galaxy. While it makes intuitive sense that such a gravitational disturbance could "wake the dragon" and trigger powerful activity, a connection has proven elusive to demonstrate statistically.

Radio galaxies, in contrast, favor elliptical galaxies, or close relatives. Many radio galaxies lie in quite normal-looking ellipticals, with some to be found in galaxies which look like ellipticals that have recently digested some kind of gas-rich galaxy. Radio galaxies are statistically more likely to have close neighbors or slight disturbances from a perfectly elliptical form than are ellipticals without radio emission, leading to a suspicion that external interactions make it easier for a galaxy to show this kind of activity. Something about ellipticals makes it easier to launch and sustain the long jets and extensive lobes of energetic particles giving rise to the radio emission. This may be the relative lack of dense gas between their stars, which gives the fast but fragile jets a clearer path to the outside than they would have in the crowded interstellar medium of spiral galaxies.

Detailed images, particularly with the Hubble Space Telescope, have shown that many, perhaps most, radio galaxies show unusual dust structures, which are usually good tracers of interstellar gas as well. It is common to see a thin dusty disk around the nucleus, perpendicular to the radio jets. This is in the right orientation for the expected accretion disks, but these disks have sizes of tens of light-years rather than a light-month. They might be connected - the accretion disk is being fed from _somewhere_, after all - but are not yet the hot, active accretion disk that we suspect produces the fireworks. If it turns out that these dust features are really more common among radio galaxies than among ordinary ellipticals, the difference may be a signal that many radio galaxies have recently acquired large amounts of gas in hostile mergers and acquisitions.

As we look at more and more distant radio galaxies, seen farther and farther back in cosmic time, they start to look more peculiar and less like the familiar, symmetric ellipticals we see here and now. At high redshifts, we see radio galaxies aligned with the radio sources, often showing very patchy and irregular forms. This empirical difference probably combines several effects. In ordinary visible-light imaging, we are seeing light which left the galaxy at shorter wavelengths when we look at large redshifts; eventually we see what started out as ultraviolet radiation, which will be strongly affected by any population of young stars or scattering by dust grains. This can be remedied, if imperfectly at present, by observing in the infrared, so we measure the same piece of the galaxies' starlight spectrum both near and far. On top of this shift, many of the radio galaxies we see at high redshift are more powerful, and hence more violently active, than any we see in the local Universe. If we have identified, say, the ten most powerful radio galaxies in the observable Universe, the nearest will most likely still be five billion light-years away, so that we cannot tell whether its properties are due to some evolution with cosmic time or the peculiarities of being such a rare and luminous galaxy. Finally, the most interesting possibility, one that we can test with larger numbers of more modest radio galaxies, is that we are seeing genuine evolution of radio galaxies with cosmic time, perhaps as they slowly settle into today's smooth and symmetric forms after a tumultuous youth in which galaxy collisions and formation of stars as the radio jets collide with gas played important parts.

But what of host galaxies around quasars themselves? When observed with the most precise ground-based techniques, a handful of bona fide quasars had shown the kind of "fuzz" that would be expected for typical galaxies. Proving this to be starlight was a formidable challenge, since the light was faint and located very close to the enormously bright quasar light. Spectra of a handful of these "fuzzballs" have been obtained, with a variety of results. Some show the absorption features of old stars, as in elliptical galaxies. Some show the distinct pattern of young stars, which are bright enough to outshine older populations. And some show only emission lines from gas illuminated by the quasar core, tens of thousands of light-years away. Improved imaging techniques, going into the infrared or using the very best optical observing conditions, also yielded detections of fuzz around many quasars, and in a few cases clear structures resembling spiral galaxies. But the real breakthrough to see what kinds of galaxies surround quasars was expected to come with the Hubble Space Telescope. Its crisp images, free of atmospheric blurring, should reveal even small and faint galaxies around quasar cores, and show us enough details to tell what Hubble types the galaxies are. This has indeed happened, though this project had to await the first servicing mission and corrective optics that restored the telescope's expected performance.

At the outset, we might have expected to see a distinction similar to what we find for Seyfert and radio galaxies - with radio-loud objects in elliptical galaxies, and radio-quiet objects behaving like beefed-up Seyferts and lying mostly in spirals. With the first batch of quasar images from Hubble, this simple picture became more complicated. Many quasars at redshifts z=0.1-0.5, corresponding to distances roughly 1.5-6 billion light-years, were selected for observation, with these low redshifts favored to get the best sensitivity for the surrounding galaxies. A wide variety of host galaxies showed up, confounding any simple scheme. Radio-quiet QSOs can occur in any kind of galaxies - spiral, elliptical, or merging systems. Radio-loud quasars occur in elliptical galaxies, but also in merging systems whose narrow tidal tails suggest that the original galaxies must have been spirals.

From the first batch of images obtained by John Bahcall and collaborators, it seemed that some quasars might be orphans, with no important surrounding galaxy. This would be an enormous step, and led to speculation that the black holes in quasars might predate the surrounding galaxies and in fact might be the seeds to start galaxy formation in the early Universe. However, reality may well be more prosaic. Observations using more closely tailored instrument settings (such as those by Mike Disney's group) turned up detections of small, though fairly bright, galaxies around a few of these "orphans". More telling has been the immediate environment of QSO galaxies as seen with Hubble. Over one third of the QSO host galaxies, of whatever kind, have small, close companions. This is the strongest single piece of evidence that galaxy encounters can trigger nuclear activity. These companions include many of a very particular type - galaxies that are very small, as if they represent the remnant cores of disrupted systems which started out in a more normal way. More typical galaxies could not fit so close to the quasar cores; there's simply not room. This may be telling us that particular kinds of gravitational disturbance are most effective at feeding the beast within. Also, this tells us that quasar activity in today's Universe largely happens in relatively brief episodes. A galaxy encounter this close must be brief - within a few hundred million years, the intruder will either escape or find itself absorbed by the larger galaxy. If we see a link between quasars and these companions, the quasar must appear as such for a span of time not much greater than the duration of the galaxy collision. Thus, today's quasars would be rare rejuvenations of objects which were perhaps more active in the early Universe.

And what of the high-redshift quasars, when they were much more numerous and more powerful than today? Seeing their host galaxies remains a formidable challenge even with Hubble's capabilities. Several factors conspire to make these distant galaxies all but invisible in the glare of their nuclei. Sheer distance makes their images huddle closer to the core, so the glare of the nucleus obliterates larger and larger regions. Most of their starlight will be seen in the near-infrared due to their large redshifts, which means that we will observe it with worse resolution, as the diffraction limit for even a perfect telescope becomes larger with wavelength. And finally, the expansion of the Universe makes the surface brightness we observe decline rapidly, as the fourth power of (1+z). Thus, we do not expect to see any kind of normal galaxy around the highest-redshift quasars, and typical ground-based infrared images seldom show any "fuzz" beyond about redshift z=1.

Still, arguments and expectations are poor substitutes for looking, and there have been attempts to find galaxies around high-redshift quasars with Hubble and with the new generation of large ground-based telescopes. At redshifts z=2-3, Susan Ridgway and colleagues have used Hubble with the NICMOS infrared camera to detect a few host galaxies, some of them no brighter then the hosts of nearby quasars. In fact, they are faint enough to suggest that these galaxies would have to grow significantly, either by accretion of gas or mergers with smaller galaxies, to become typical galaxies today. Looking for more extreme kinds of stars in the host galaxies, a group led by Matthew Lehnert examined the ultraviolet light from hosts in this same redshift range, finding that some show knots and clumps of light from star-forming regions. The history of star formation in the galaxies around quasars may tell us their history, particularly how the brightness of the nucleus and growth of the galaxy might be linked.

Since exceptions often tell us about the rules they defy, it is interesting to find that there may even be an exception to the strong distinction between hosts galaxies of Seyferts and radio galaxies. In a survey of hundreds of radio galaxies in clusters, a group including Frazer Owen, Michael Ledlow, and myself, turned up a powerful double-lobed radio galaxy in a highly flattened disk galaxy. If this is indeed a spiral, it's the only such case known, and presumably can teach us interesting things about just what a galaxy needs to do to entertain such long-lasting and powerful phenomena. The galaxy turns out to be exactly edge-on, as shown nicely in this recent image from the 8-meter Gemini-South telescope (courtesy of the Gemini Observatory). It's hard to get a very exact galaxy type when it is so exactly edge-on, but the dust structure is irregular enough to suggest it's really a spiral instead of an arm-free S0 galaxy. New Hubble ACS images may be showing show us clusters of young stars, quite literally resolving the matter.

My graduate advisor, Joe Miller, once mused, during a drive back and forth to Lick Observatory, that we didn't know whether quasars were merely interesting, or actually important. For many scientists, no such distinction exists, but what Joe meant was that it wasn't clear that the phenomena of active galaxies stood in the path that led to the most precious of all cosmic phenomena - life. Do they reflect the same properties of the Universe that allow us to exist, or could our existence have been substantially the same had the Universe been such as not to bring forth these powerhouses? The recent findings on how many galaxies seem to contain dead quasars suggest that the answer may be "yes" on both counts. Quasar activity may be an important part of the development of every luminous galaxy.

For many years, quasars were the only objects that we could trace to the large redshifts that tell us about early cosmic history. Thus, many cosmologists entertained fond, if implicit, hopes that quasars really were closely connected to the formation of galaxies. Now we find evidence that they really are connected to the history of galaxies, in several ways, so that there is a close connection between the bits of cosmic history we can trace with quasars and the development of galaxies (including our own). Massive black holes in galaxies, which show up as active nuclei when they are accreting material at high enough rates, are basic features of galaxies, and first occurred in very particular environments.

It has been a long-standing curiosity that quasar spectra look much the same whatever the quasars' redshifts. Nearby objects and those seen at early cosmic times show much the same intensities of emission lines, which tell us about the relative abundances of such heavy elements as oxygen, nitrogen, aluminum, and iron. These are produced in massive stars and supernovae, with somewhat different ratios depending on the mass and age of stars involved. We can now examine quasars at redshifts as great as z=6.4, which is within about 800 million years of the beginning for the most likely cosmological parameters. The gas illuminated by the central engines of these quasars is no less rich in the products of stellar fusion than what we see here and now. In fact, if anything, the products of the most massive stars are relatively more common in these early quasars than today. Fred Hamann and collaborators have shown that this situation makes sense if quasars reside in regions that are special, even beyond the uniqueness of possessing supermassive black holes. They must also have hosted early and intense bursts of star formation. These must have been intense to have produced the masses of heavy elements seen, and early in order for supernovae to have already produced the heaviest of these elements and blown them back into interstellar space where they form part of the material seen as the quasar's emission lines. In fact, certain elements are notable by their relative rarity, which gives a further clue. Type Ia supernovae (the kind useful for cosmology because their luminosities are so nearby the same) produce much greater amounts of atomic nuclei near iron in the periodic table than either fusion in the cores of red giants or the type II supernovae that result from the collapse of massive stars at the end of their lifetimes. These elements are relatively deficient in high-redshift quasars, indicating that, while star formation and deaths of massive stars has been brisk, type Ia supernovae have not played a role. Since type Ia supernovae are thought to result from mergers of binaries containing white dwarfs, a process with a time scale upwards of a billion years, this makes sense. No matter how brisk the star formation might have been for several hundred million years, no Type Ia supernovae and their iron-peak excess would appear for about a billion years. Still, there is a serious issue in just how early such star formation could have begun, tied up with the history of galaxy formation.

All this seems most palatable if the presence of an intense burst of star formation and the enormous central black hole are somehow connected, either with one causing the other or both as byproducts of events as the surrounding galaxy takes shape. Indeed, a strong connection between central black holes and the surrounding galaxies has emerged as we developed the techniques to find quiescent black holes. Unlike the flamboyant active nuclei, which advertise the central engine everywhere from radio waves to hard X-rays, black holes that are quiet can be found only be their gravitational effects on surrounding objects. In the same way as an ordinary companion star can betray the existence of a black hole formed from a single star, the motions of stars near a galaxy's center can tell us of the presence of the kind of concentrated mass that we interpret as a supermassive black hole. The earliest hints of such a monster came from observations of the radio galaxy M87 in Virgo, presented in by Peter Young, Wallace Sargent, and their collaborators, including one of the earliest applications of then-new charge-coupled devices (CCDs) to astronomical research. They found that the stars very near the core of M87, in which the radio emission and powerful jet show ample evidence of unusual activity, are more strongly concentrated than in other elliptical galaxies, and that their typical speeds increased so rapidly towards this core that no amount of normal matter could account for the required gravity without being visible. They thus proposed that the nucleus of M87 contains a black hole of about 5 billion solar masses. (Further observations, including the motions within a disk of gas revealed most clearly by HST observations, have strengthened this conclusion.)

The M87 observations pushed the state of the art, and such measurements have remained at the limit of what one could do from the Earth's surface. Definitive evidence of black holes in galactic nuclei had to wait for the Hubble Space Telescope, and in particular for the installation of the Space Telescope Imaging Spectrograph (STIS) during the second servicing mission in early 1997. This instrument could measure the spectra of galaxies, not only with higher efficiency than its predecessors, but two-dimensionally, measuring a slice through a galaxy all at once rather than one point at a time. STIS has proven its value as a black-hole meter par excellence, delivering so many measurements of black holes in galaxies that we can now examine their demographics instead of their mere existence. It is important to do these measurements using the motions of stars rather than gas, even though the emission spectrum of ionized gas is much easier to measure. Interstellar gas can be shoved around by many things other than gravity - magnetic fields, shock waves from supernovae, and heating by ultraviolet radiation can play roles. Stars, in comparison, act almost as inert particles in tracing the galactic gravitational field.

In our own Galaxy, it has now become possible to do this experiment with stars, and do it one-by-one rather than statistically. Speckle interferometry, and later adaptive optics, have revealed stars in the innermost few light-years around the nucleus of the Milky Way, and shown their locations so accurately that their orbital motions are manifested over only a few years. Beautiful results by Andrea Ghez and collaborators at Keck and by Reinhard Genzel's group using the 3.5-meter New Technology Telescope at La Silla, Chile, and more recently using adaptive optics at the 8-meter VLT, make it clear that stars near the galactic center are affected by an invisible object of about one million solar masses, located right where the tiny central radio source appears. In the same way as Master Yoda and his disciples saw through an attempt to wipe a planet from the Jedi archives, we can discern the existence of this object through its gravitational signature.

Our galaxy's central object proves to be quite modest by the standards of many other galaxies we can observe. In these more distant cases, we use the amount of broadening of the spectral lines due to the different Doppler shifts of stars seen in each region to map the mass. As shown most clearly by John Magorrian and co-workers, the HST data show a surprising relation between the mass of the central object and the properties of the surrounding galaxy. In virtually every case, the black hole comprises about 0.5% of the mass of the stars in the spheroid of the galaxy. This is the roundish central bulge of spirals, or the whole galaxy for ellipticals, and in each case is thought to represent the results of a rapid early epoch of star formation. So either the growing galaxy knew about the central black hole, or the black hole knew about the surrounding galaxy's stars, in some way that allowed one to control the other.

This so-called Magorrian relation might mean either that the black holes were there first, and always attracted a closely proportional amount of material to make stars, or that the galaxy was there first and always grew a central black hole to a constant fraction of the stellar mass. There is such a clear theoretical explanation of the second alternative that it seems vastly more likely than the "magic" needed if the black holes were there first. If we imagine a galaxy with a modest central black hole, say one made by merging only a single cluster's worth of stars, the black hole initially has little dynamical impact on its surroundings. As it grows, however, it eventually comes to dominate its immediate surroundings gravitationally, and alter the orbits of surrounding stars. Once the black hole reaches a particular maximum mass, according to calculations by David Merritt and Gerald Quinlan, it shuts down its own growth, by forcing the orbits of nearby stars to be more circular and thus keep them safely out of its reach. Their calculations give results that track the Magorrian relation in an interesting way, although they do show the black hole shutting its growth down at a mass fraction several times higher than the 0.5% typical of bright galaxies. If this is what happened in the lives of most galaxies, as suggested by the prevalence of quiet central masses in most of the galaxies so far observed with STIS, then quasars may indeed by important as well as interesting. Most galaxies must have undergone a bright quasar-like phase when the black hole was growing, ingesting material and surrounded by gas to produce the quasar symptoms. The presence of the central black hole may well be an unavoidable part of a galaxy's formation and evolution.

The statistics of quasars can tell us how long this took, and when most of the action happened in the cosmic calendar. The census is easiest using X-rays, for several reasons. Quasars have fewer other kinds of starlike objects to hide among in X-rays, compared to visible light or the infrared. X-rays of enough energy are not blocked by intervening dust or gas, so we are not biased by these in a quasar's environment. The very deep and very sharp X-ray images from the Chandra X-ray Observatory have allowed Amy Barger and colleagues to ask when most of the X-rays from quasars began their journey toward our detectors, and thus when massive black holes were growing and radiating most powerfully. Combining the X-ray intensities with redshifts measured for many of the objects in the optical, they show that the typical active nucleus has spent about 500 million years (or 5% of its lifetime) "on". Much of this must have happened in the early Universe to fit what we know of the populations of quasars at various redshifts. This fraction of 5% matches the fraction of bright galaxies that host Seyfert nuclei today, although there must be a wide range in the "duty cycles" of various galaxies. It has long been clear that activity must turn on and off, simply because keeping the most powerful quasars going continuously for the age of the Universe would consume entire galaxies and leave more massive black holes than we see anywhere. This also makes sense with the observation that many quasars in the local Universe have close interacting companions, which, as noted above, suggests that quasar activity occurs in relatively short episodes.

Quasar hosts at high redshift must have been very special places - the first galaxies to pull themselves together enough to begin forming stars in a dense core, the same ones that could grow a central black hole most rapidly. Even with numerical models showing how the growth of a central black hole can interact with the dynamics of the surrounding stars, there remains a considerable mystery about original formation and early growth of these black holes. They could grow rapidly, by disrupting passing stars and accreting some of their matter, only when already massive enough to exercise strong tidal forces across interstellar distances. How did they grow so massive, in the relatively short times available when we see the earliest quasars? This might be understood if the black holes started out that way, some kind of primordial relics of a chaotic time in the early Universe, perhaps the mass concentrations that seeded and accelerated the formation of galaxies. This notion was popular for several years, when it seemed that Hubble images showed some quasars with little or no surrounding galaxy. However, as further observations showed that all nearby quasars do have a substantial surrounding galaxy, and in the light of the tight correspondence between the mass of the central object and the surrounding stellar spheroid, it appears more likely that the black hole has grown along with the surrounding galaxy's development. Thus we are driven to look elsewhere for a way to bridge the gap from black holes made by single collapsing stars, and ones powerful enough to shred other passing stars for breakfast.

The most massive black hole that we can understand from the collapse of a single star would have perhaps 100 solar masses, and that only if the star had been one of the first generation of stars in the Universe. (Because the final explosions of such stars are so powerful, we actually don't expect to find their remnants within today's galaxies). This falls well short of the 100,000 or so needed to start accreting mass from other stars, which form a much richer fuelling source than gas at the centers of most galaxies. One way that this gap might be bridged lies in the properties of relativistic star clusters. These are clusters which are so compact that we need relativity to properly describe the motions and interactions of their constituent stars. If we imagine taking a rich present-day star cluster, turn its members into neutron stars or black holes, and squeeze the whole group into a volume a couple of light-years across, that starts to become a relativistic cluster. Forming such a cluster would have to start with an extraordinarily intense and compact burst of star formation, followed by dynamical interactions once some of the member stars had ended their lives by becoming neutron stars or black holes. The most massive objects will tend to settle to the center in the ordinary course of interactions among stars, a process which accelerates as the cluster becomes relativistic. In fact, a long-standing theoretical problem has been the stability of relativistic star clusters. Is it inevitable that such a cluster will fall together until its central members form one enormous black hole, or can such a cluster last for long cosmic times? There is at this point no clear-cut verdict; for some starting conditions, the cluster will collapse, while others appear to be long-lasting. The reason for even considering such extreme (and, at this point, unobserved) assemblages is that we do see regions of very intense star formation, loaded with massive stars, in nearby galaxies that most nearly mimic our pictures of nascent galaxies at early times, and this kind of environment also matches what we need to fit the chemical makeup of the gas in quasars. Still, this is an area ripe for further study, and more data on conditions in early galaxies will certainly help.

The "standard" model of the central powerhouse in an active nucleus features a very massive black hole surrounded by an accretion disk, and it is in fact that accretion disk to which we might attribute most of the radiation we can actually see. But, to this point, the accretion disk has proven very shy, and observational tests for direct signatures of the disk have come up negative or ambiguous. This has held true with features in the optical spectrum, the overall shape of the ultraviolet spectrum, and lines from very hot matter seen in the X-ray spectral region. Searches for some feature that comes directly from the accretion disk rely either on the characteristic pattern of motions in a thin, orbiting disk, or on that fact that it will be heated by internal collisions and will thus be hot - like a large, hot, and very strangely shaped star.

The profiles of emission line in the spectra of most AGN are centrally peaked, telling us that only minor fractions of this light come from material at very high velocities. This sounds more like an amorphous cloud, or maybe a spherical distribution of clouds, than a thin, organized disk. Thus there was excitement when a class of quasars and radio galaxies turned up showing "twin peaks" - double emission-line profiles, with most of the radiation either redshifted or blueshifted compared to the center. Generically, this is just what we expect from a disk, and it is also just what we see from the hot disks around white dwarfs or neutron stars that are accreting gas from a companion. The first to be examined in detail was Arp 102B, part of an interacting galaxy pair that first seemed interesting based on an X-ray detection. A dedicated survey by Michael Eracleous and Jules Halpern turned up more of these objects, suggesting that they comprise nearly 10% of radio-loud objects if you look carefully enough. The double-peaked profiles did seem to show some of the right features to represent light coming directly from the accretion disk. In particular, the blue peak was often a little sharper and brighter than the red peak, effects predicted from relativity as the light from the approaching side is boosted in our frame of reference. Life was not to remain so simple, though - it turned out that the red and blue peaks in some of these spectra can vary independently, whereas the two halves of an accretion disk should remain coupled to each other within hours as material follows its independent orbits.

Instead of emission lines, the accretion disk might well emit only continuum radiation in the optical range. If so, its signature might be such a broad spectral pattern that it would show up only when comparing data ranging from the ultraviolet into the infrared. Matt Malkan and Wallace Sargent carried out a widely-quoted study of this kind, showing that the broad emission pattern known as the Big Blue Bump does look like the kind of distorted stellar atmosphere that an accretion disk could give. It also showed some of the kinds of changes from one object to another that would result from effects of relativity as the central mass and viewing angle changed, according to calculations by Wei-Hsin Sun. So far so good. But the idea didn't pass other tests that these results suggested. In particular, the light output of such an object (including the surface of a supergiant star, which we understand rather better) should change dramatically at the Lyman limit in the ultraviolet, where radiation on the blue side of the limit can ionize hydrogen atoms. This change would be either a step in intensity or a change in polarization, unless the disk were carefully constructed to cancel these signals. Robert Antonucci and several sets of collaborators have tested this ideas, and find that they do not correspond with what we actually see in quasars and radio galaxies. As they suggest in a recent paper, if such quasars do exist, they are the ones we can't observe in the ultraviolet because the outer parts of the disk are so broad that they block our view. This is not only a suspiciously neat coincidence, but it violated the very small number of objects as powerful as quasars but showing only narrow emission lines (as if we don't see the core region at all).

The exploration of detailed spectra of X-rays from active nuclei has become a rich field, first with the Japanese ASCA satellite, and more recently with NASA's Chandra and ESA's XMM-Newton observatories. These can now show us structures in the velocity of hot X-ray emitting matter that we could formerly find only if it also gave off emission lines in the optical or ultraviolet ranges. Some of these emission lines, now from gas so highly ionized that it must be at temperatures of tens of millions of degrees, showed the kinds of velocity structure that we might expect for a disk. Some even showed what looked like distinct double peaks. But, again, more data have clouded the situation. There are several distinct contributions to the X-ray spectrum in the crucial energy ranges, so that it is not completely clear how much of a broad bump comes from a particular spectral line and therefore what its velocity structure is. And in some cases, the apparent double structure has gone away as better data accumulated. The answer to this one should become clear, as Chandra and XMM-Newton continue observations and the community becomes more experienced at their interpretation.

Disklike structures have indeed been seen in many nearby active galaxies, radio galaxies and Seyferts alike. These are most often seen as dark features from dust absorption in Hubble images, and span diameters of 50-500 light-years. In many radio galaxies, these disks have just the orientation we would expect - closely perpendicular to the radio jets. There is no doubt that these are disks, and that they are related to the central activity, but they are not "the" accretion disks in these objects as postulated in the standard scheme. These disks are much too large and much too cold. The active accretion disk must be hot precisely because it is so dense, and so close to the central mass, that orbital energy is shed outwards into heat, allowing parcels of gas to fall ever closer to the core until no stable orbit is possible, and it vanishes through the black hole's event horizon.

Some theorists have risen to the challenge posed by recent observations, by noting that a significant level of accretion could happen more quietly - a situation known in the jargon as advection-dominated accretion. If this is happening, a disk geometry may exist farther from the core, as we often see in dust or gas disks, but break down close to the central black hole. The energy release would still occur in this volume, of course, and there may be hope of testing this idea's predictions as coordinated observations of quasars from the gamma-ray to radio windows become more precise and more common. Still, the notion of an inner disk makes a great deal of sense in accounting for the radio jets in active nuclei. There is hope that upcoming instruments using optical interferometry - such as NASA's Space Interferometry Mission - would be able to detect disks by their shapes, independent of any particular kind of spectral signature.

We may be missing traces of the central engines because we lump together quite dissimilar objects. Active nuclei span a wide range in power, radio/optical ratio, and dynamics as judged by their spectral lines. How many kinds of objects are there? Investigators have tried for years to produce the AGN counterpart of the Hertzsprung-Russell diagram for stars - a simple plot, involving straightforward quantities, that separates physically different kinds of objects (such as red dwarf and red supergiant stars), leading to a comprehensive theory as successful as our understanding of stellar evolution has become. Some groupings have emerged by considering such quantities as the relative strength of the X-rays, whether an object is a strong radio source, and how symmetric the emission line profiles are. Attempts by Tod Boroson at the National Optical Astronomy Observatory, and a multinational conglomerate headed by Jack Sulentic, now suggest that key features in how a quasar appears may be the central black-hole mass and the rate at which it is accreting mass. There may be a critical accretion rate at which the structure of the whole emitting region changes. Oddly enough, these properties don't seem to correlate in a way that fits the otherwise well-supported notion that much of what we see in quasars depends on our viewing direction with respect to the disk. This is a surprise that is still percolating through our mental picture.

A recurring theme in quasar astronomy has been the quest to find them at ever-higher redshifts. This goes beyond the desire to have a record, however fleeting, and the associated bragging rights. Very distant objects tell us interesting things about cosmic history, first by their very existence, and in addition by what the reveal about the otherwise invisible Universe in front of them.

It has long been clear that the quasars we can observe evolve strongly with cosmic time, having gone through a peak about 12 billion years ago when a large fraction of bright galaxies must have hosted quasars, and with their number falling dramatically down to the present day. Going outward in redshift and farther back in time, the number of quasars (at least the very powerful ones that we could find so far away) declines above redshift z=4, with progressively fewer and fewer up to the current record at z=6.4. Either we are watching the population of quasars "turn on", or for some reason we can't find most of the high-redshift quasars that were there.

As quasars at higher redshifts are found, their existence places more and more stringent constraints on how rapidly such compact objects formed in the early Universe. The Sloan Digital Sky Survey has yielded quasars at z=6.4, only about 800 million years after the beginning in the most popular cosmological scheme. In this time (during which our Galaxy would rotate only three and a half times at the Sun's distance) at least a few galaxies must have formed, collected enough material to make a fairly massive black hole, and had enough formation of massive stars in its immediate environs to produce the heavy elements seen in quasar spectra. For example, finding quasars at redshifts so high that they are seen at half this cosmic age would be a real problem, showing that there was something very basic about early cosmic history that we don't understand (ignoring for a moment the high probability that there is indeed something important about galaxy and quasar formation that we are blissfully unaware of not understanding). The Sloan survey will be able to find very luminous quasars to about z=7, and less luminous but more common objects should turn up in deep infrared surveys. We would also be alerted to QSOs at very high redshift by their X-ray emission, a helpful addition to the toolkit since absorption by interstellar gas is strong only for those X-rays that we already lose because of the objects' redshift.

Given that these quasars do exist, at least in small numbers, at high redshifts, we can make extensive use of them to study the material in front of them through its effects on the quasar light, even if the foreground material is otherwise invisible. This makes use of the absorption by foreground gas, virtually the only way to trace the tenuous gas between galaxies, and gravitational lensing by any kind of mass concentration in front of the quasar.

Gas in front of any bright background source will absorb certain, very narrow, wavelength ranges of the background light. The wavelengths depend on the chemical mix, density, and temperature of the gas, and we will observe them shifted by any relative motion between the absorbing gas and ourselves. With quasars as the background lighthouses, we see gas in distant galaxies close to our line of sight (galaxies that are often too dim to detect directly) and the widespread gas between galaxies - the intergalactic medium. As we look across cosmological distances, various clumps of gas will appear at different wavelengths because the quasar's light encountered them at different redshifts, allowing us to map this tenuous material in space and time. This is how we came by virtually everything we know about the intergalactic medium.

As soon as quasars could be observed at redshifts above about z=2, a new spectral feature became apparent - a whole series of very narrow absorption lines, appearing blueward of every quasar's own Lyman alpha emission line. Lyman alpha is a very special spectral line, the strongest transition of hydrogen, itself the most abundant chemical element. The match of these new features with Lyman alpha was too close to be coincidence, leading Roger Lynds in 1971 to identify these absorption features as the same line, seen at smaller redshifts from material in the foreground. The lines would naturally cease at the quasar's redshift, since more distant material would be behind the quasar and not affecting its light. The "Lyman alpha forest" appears in the spectrum of every high-redshift quasar, and the statistics of these features have been mapped in great detail. For almost two decades, this industry proceeded with the goal of connecting these clouds to galaxy formation, perhaps with the Lyman alpha clouds representing pre-galactic gas which was condensing over time to make galaxies or being accreted into galaxies. This certainly fit with the number of clouds seen at various redshifts - the density of Lyman alpha absorption features increases dramatically with redshift, which means that they are vanishing rapidly with cosmic time. It was a bit of a surprise to see, from early Hubble Space Telescope spectra in the deep ultraviolet, that a few of these absorbers survive to the present day, revealed only when space telescopes could look for them at small redshift, and thus still in the ultraviolet, at the kinds of sensitivity that has long been available using large ground-based telescopes at high redshifts.

Two other things were known from the Lyman alpha forest. One was that the gas we can see is only a small fraction of what must be there. To produce Lyman alpha absorption, or indeed absorption in any of its possible spectral lines, hydrogen atoms must be neutral - that is, in their ordinary state including an electron. Ionized hydrogen, in which the electrons have been separated from the nuclei, will not produce absorption lines. Comparison of the small widths of the absorption lines, indicating how large the internal motions can be in the absorbing gas, with the amount of gas needed to produce the absorption we actually see, tells us that the gas we measure doesn't have enough gravity to hold itself together. Furthermore, at redshifts and thus locations near quasars, we observe a near-complete disappearance of the Lyman alpha forest. This fits in the same picture, since the quasars are copious sources of ultraviolet and X-ray emission which can ionize such rarefied gas for millions of light-years around. So we had a general picture of hot, mostly ionized clouds of gas floating between the visible galaxies, vanishing over time as they either formed into galaxies or were acquired by existing galaxies.

This scheme underwent a complete overhaul in the 1990s, one which almost wiped out some of the traditional goals of studying the Lyman alpha forest and presented us with a much richer and more complete view. A series of numerical simulations of intergalactic gas, especially work presented by Renyue Cen at Princeton, demonstrated that a pair of biases resulting from the behavior of patchy gas in an expanding Universe could produce what we see without the gas being held together in small, distinct clouds, but instead resulting from the same pattern of large-scale structure that encompasses clusters and superclusters of galaxies.

Efforts to understand the formation of galaxies and clusters have included simulations of how matter behaves at the temperatures and densities encountered in the early Universe, starting from the tiny fluctuations derived from the cosmic microwave background and watching what happens as the Universe expands, the matter cools, and the state of the cooling gas changes. The results have been described as a cosmic web, with the densest material, most likely to begin galaxy formation, tracing sheets and long filaments through space. The properties of the Lyman alpha forest can be understood beautifully as part of this same structure, made to appear even more clumpy and spiky than it really is by two amplifying effects. First, even if most of the gas is very hot and ionized, the fraction that is neutral, and thus shows up in absorption lines, is much higher in denser regions. This comes about because electrons and atomic nuclei are more likely to collide and form neutral atoms (in a process known as recombination) where there are more of each. In fact, the rate of this process increases as the square of the gas density, so that the amplitude of fluctuations in Lyman alpha absorption would look like the square of the fluctuations in the actual gas density from this process alone. Moreover, the structure in the gas is not static - it is influenced by gravity, which will draw material into denser regions and further evacuate the emptiest areas. This motion means that, as we look through a sheet or filament traced by the absorbing gas, it will appear narrower in redshift than we would expect for its actual depth in space. The gas on the front side will be falling in away from us, so its redshift is a little too high for its actual position; while the opposite is true for material on the far side. Putting these effects together, Cen and others have shown convincingly that the Lyman alpha forest is the observed guise of a richer, and much more pervasive, web of gas pervading the Universe than we thought only a decade ago.

Hard work, and new instruments, have turned up additional species that trace this intergalactic web. Helium has an absorption line which corresponds precisely to Lyman alpha, produced by helium which has been ionized so that only one of its two electrons remains. This line falls even further into the ultraviolet than the hydrogen line, so it can be observed only from high-redshift quasars and then only deep in the ultraviolet. This is because there is so much cool hydrogen in most galaxies, including our own, that radiation in a broad swath of the ultraviolet spectrum is completely absorbed in ionizing some of that hydrogen. This produces a piece of the spectrum shortward of 912 Angstroms (known as the Lyman limit) where we cannot see very far through our Galaxy, and cannot see outside it at all. Thus, to see so deep in a quasar's ultraviolet spectrum, we must see it at such a high redshift that its helium Lyman-alpha line, starting at 304 Angstroms, enters the Milky Way at 912 Angstroms or longer. To make this measurement more challenging, the quasar's light also cannot pass through any other galaxy's neighborhood along the way, since the hydrogen in that galaxy will absorb the ultraviolet radiation. Finding this absorption - known as the He II Lyman alpha absorption - has been a major project, connected with the history of heating of the intergalactic gas as well as its structure. It has involved extensive work with the Hubble Space Telescope, the shuttle-borne HUT experiment, and the FUSE satellite, supported by spectra of the matching hydrogen lines from the largest ground-based telescopes. This phase of helium, which can be ionized only by radiation four times more energetic than needed for hydrogen, is plentiful in the intergalactic realm. Despite that fact that helium is generally only about 3/10 as common as hydrogen, the He II forest is stronger than that seen in hydrogen; some weak lines can be detected only in helium. This is because a larger fraction of the helium is observable than for hydrogen, since a large fraction of the hydrogen is ionized and does not produce absorption lines. The He II feature also occurs more frequently in less-dense regions, farther from the major concentrations of gas, where small levels of radiation from quasars or galaxies can keep the gas more completely ionized.

Adding to the excitement was the discovery that, as well as the hydrogen and helium that were left over from the first few minutes of cosmic history, the intergalactic gas includes elements which must have been synthesized in massive stars and scattered in their final supernova explosions, providing evidence of an early and widespread episode of starbirth. Key evidence came, again, from quasar spectra. Using Hubble Space Telescope spectra, a group led by Todd Tripp found that there is another absorption-line forest, this one from O VI (that is, oxygen which has been stripped of 6 of its 8 electrons), seen from a pair of absorption lines in the deep ultraviolet. There is not yet a good estimate of just how much mass these lines arise from, since additional oxygen could be hiding with either more or fewer electrons, and it is not clear how much of other elements such as carbon and nitrogen might also be present. However, even conservative assumptions on these issues suggest that the amount of such processed gas between galaxies may be comparable to what we find in the galaxies' stars themselves - a significant reservoir indeed. In the light of simulations of the formation of the first stars as well as galaxies, we see hints not only of what the intergalactic medium is like, but how this all-but-invisible player has interacted with the galaxies we see.

The first stars, forming from pure hydrogen and helium, would be quite different from the ones we see around us today, which began with a small but important salting of heavy elements synthesized in the cores of earlier stars. The first stars (sometimes called "population III" by analogy with the two populations of stars we see today) would have been very massive and hot, exploding with more violence than today's supernovae. These stars, with hundreds of solar masses, would have scattered their yield of heavier elements thinly but widely as each exploded, destroying the surrounding gas clouds in the process and forcing galaxy formation to start over, but now with star formation proceeding as we still see it thanks to the cooling effects of heavier elements. And we conclude all this from narrow bands of missing light in quasar spectra.

Looking for broader gaps in quasar spectra can also tell us important things, now about the cosmic history of mass and energy on the largest scales. In standard cosmology, the cosmic microwave background represents the radiation travelling at the time when the overall gas became neutral and hence transparent at most wavelengths, so that this radiation was free to travel. We still see some of it today, redshifted by a factor of more than a thousand. However, something else must have happened to the intergalactic gas since that time - from the Lyman alpha, He II, and O VI forests of absorption lines, we see that intergalactic gas is mostly ionized today and has been since the time corresponding to at least redshift z=5. What else has happened?

James Gunn and Bruce Peterson pointed out in 1965 that any important amount of gas in a cool, neutral form would absorb a large "trough" against quasar spectra, since smoothly distributed gas would occur at all redshifts. It soon became clear that we do not see such Gunn-Peterson troughs in available quasar spectra, something that remained true until quite recently. As we have seen, what gas there is in intergalactic space is very patchy (producing narrow absorption lines instead of broad troughs) and mostly ionized (so that the absorption is much weaker). Gunn-Peterson absorption has finally been seen in the spectra of the very highest-redshift quasars known, found by the Sloan Digital Sky Survey (see the technical journal paper). The density of neutral intergalactic gas declined dramatically from z=6.3 to z=5.8 (corresponding to a time interval of about 100 million years, very short for such a universal transition) and has been unobservable since that time. We may finally be seeing the tail end of the epoch of reionization, the event that ionized the cosmic gas and has kept it so since. The most important energy source, based on the amount of energy required, seems to have been the initial birth of ordinary stars in galaxies - not the rare, if brilliant, population III stars, but the most massive and hottest stars formed when today's galaxies began their careers. Since then, the small fraction of quasar radiation has become more important in maintaining the ionization of intergalactic gas. As the Universe expands, the gas requires less radiation to keep it ionized, as electrons and atomic nuclei encounter one another less often, and the overall rate of star formation in galaxies has declined dramatically for several billion years. Once again, quasars have provided important clues to otherwise invisible aspects of galaxy history. This graphic (right), courtesy of George Djorgovski at Caltech, illustrates the Gunn-Peterson effect as seen in the spectrum of a high-redshift quasar from the Sloan survey.

Finally, quasars have been important probes of mass concentrations, seen or unseen, in the distant Universe, through gravitational lensing. One of the striking immediate successes of Einstein's general theory of relativity was the observation of the "bending" of light paths by a massive object (such as the Sun). The light from a distant quasar can find itself similarly redirected, by the curvature of spacetime, on passing such massive objects as galaxies or clusters of galaxies. We see this effect, when it is large enough, as a splitting of the quasar into multiple images on different sides of the massive "lens" The first clear example of this was the quasar 0957+561 in Ursa Major. The two images show virtually identical spectra, an important fact in deciding whether such a pair is lensed or an actual double quasar. By now, about 50 such lensed quasars have been established (see the WWW compendium run by the CASTLES project). Analysis of their statistics can tell us such things as how large the cosmological constant might be, and how many clusters or galaxies are much fainter than the average we've come to expect. Such lenses can also be used to measure the Hubble constant independent of such traditional techniques as Cepheid variable stars, when the quasar is variable enough that we can see the same variations in both images and trace the time delay between images, caused by the different path lengths for light on the two paths. So far, the error bars are no better than for other techniques, but it is a powerful confirmation that our knowledge of the Hubble constant can't be, say, wrong by a factor of 2, since such utterly different techniques show reasonable agreement.

Quasars and their kin have proven to be interesting, in ways from testing relativity through probing intergalactic gas and tracing the history of galaxies. There are signs that they may even be important, and that the next 40 years of our exploration will yield a yet richer view than we can conceive of now.

(ostentatious throat-clearing) "The views, opinions, and conclusions expressed in this page are those of the author and not necessarily those of The University of Alabama or its officers and trustees. The content of this page has not been reviewed or approved by The University of Alabama, and the author is solely responsible for its content. "

Last changes: August 2003