Components of the universe

PLANETARY SYSTEMS

The Sun.

The observed rate of release of radiant energy by the Sun equals 3.86 10³³ erg/sec (ergs per second). The particles of radiation (photons) stream more or less freely from a layer called the photosphere, which in the Sun is at a temperature of about 5,800 K (kelvins; 5,500 C or 10,000 F). The distribution of wavelengths is characteristic of a thermal body radiating at such a temperature; therefore, in accordance with Planck's law, it peaks in the yellow part of the visible spectrum. The solar luminosity is enormous, but it is much less than it would be if the photons in the hot interior of the Sun could also stream freely. However, the high opacity of the material regulates the actual outward progress of the photons to a slow stately diffusion. Indeed, the blockage of diffusive heat is so severe in the envelope of the Sun that its layers are unstable to the development of convection currents, which gives the atmosphere of the Sun a granular appearance.

The observed radius of the Sun equals 6.95 10¹⁰ cm and is understood to be the result of a balance of forces between the Sun's self-gravity and the pressure of its hot gases, which exist in a nearly fully ionized state (a plasma of positive ions and free electrons) in the deep interior. The plasma in the core of the Sun is compressed to temperatures (about 1.5 10⁷ K) that are sufficient to provide a rate of thermonuclear reactions that just offsets the slow diffusive loss of radiative heat. Thus, the Sun constitutes a controlled fusion reactor capable of sustaining its present steady loss of radiant energy for a full 9 10⁹ years before all of its initial supply of hydrogen fuel in the core has been converted into helium. From the radioactive dating of meteorites, it has been estimated that the solar system is 4.6 10⁹ years old. If this is the age of the Sun, then it is roughly midway through the phase of stable core hydrogen fusion--i.e., the "main-sequence" phase of stellar evolution. 

The Sun is too opaque to electromagnetic radiation to allow a direct look at the nuclear reactions inferred to take place in its interior. Weakly interacting particles called neutrinos offer a better probe of such reactions because they fly relatively freely from the centre of the Sun. Attempts to measure solar neutrinos by means of radioactive chlorine techniques have found levels that are only about one-third the best theoretical predictions. One possible explanation supposes that neutrinos possess mass and can be converted to (oscillating) forms undetectable by conventional schemes during their passage through the dense solar plasma. Unfortunately, experiments using purified water or large amounts of gallium as the detecting medium have contributed conflicting data with respect to this interpretation.

An indirect line of evidence suggests that the source of the discrepancy may lie more with unknown neutrino physics than with uncertain solar models. Precise measurements of the small oscillations of the solar surface induced presumably by motions in the convection zone allow astronomers to study the properties of waves propagating through the Sun's interior in an analogous fashion to how earthquakes allow geologists to study the properties of the Earth's interior. These investigations reveal that the Sun behaves similarly, though not exactly, as the best theoretical solar models predict. They also show the Sun's radiative core to rotate at about the same angular speed as the mid-latitudes of the solar surface, too slow to have any of the anomalous mechanical or thermal effects that have sometimes been hypothesized for it.

The outermost layer of the Sun turns once every 25 days at the equator, once every 35 days at the poles. This differential rotation may couple with the Sun's convection zone to produce a dynamo action that amplifies magnetic fields. The basic idea is that magnetic fields carried upward (or downward) by convection currents are twisted and amplified by the differential rotation. "Ropes" of high field strength buoy to the surface where they pop out as loops into the corona of the Sun. The corona is an extended region containing very rarefied gas that lies above the photosphere and a transition region called the chromosphere; the temperature of the corona is about 2 10⁶ K. The anchor points of the ropes of high magnetic flux in the photosphere correspond to sunspots, regions where the gas is cooler than the average photospheric temperature of 5,800 K. Thus, these spots appear relatively dark against the bright yellow background of the general photosphere.

Sunspots appear, migrate about the solar surface, and disappear as the plasma to which they are anchored moves under the influence of rotation and convection. The average number of sunspots increases and decreases more or less regularly in an 11-year cycle; however, there have been prolonged minima in history. It has been proposed that these prolonged minima correlate with changing climate conditions on the Earth, although the precise mechanisms for effecting such changes remain unclear.

Other manifestations of magnetic activity arise because of the motion of the flux ropes. It is believed that flares occur on those occasions when two flux ropes of opposite polarity are pressed against each other, and the opposing magnetic fields annihilate in a catastrophic event of magnetic reconnection. The energy stored in the field is thought to go into accelerating fast particles (solar cosmic rays) and into heating the ambient gas, which, being rarefied, has very little heat capacity. Magnetic activity of this type may be what maintains the corona at much higher temperatures than the photosphere.

Pictures of the solar corona taken during the U.S.-manned Skylab missions (1973) showed that hot coronal gas trapped in closed loops of field lines becomes dense enough to emit appreciable amounts of X rays. In contrast, coronal holes lacking X-ray emission correspond to regions where the magnetic field is too weak to keep the gas trapped and the hot gas has burst open the magnetic-field configuration, expanding away from the surface of the Sun as part of a general solar wind. 

The presence of a solar wind blowing through interplanetary space was first deduced from observations made during the 1950s of the ion tails of comets. With the advent of Earth-orbiting satellites, the particles and fields carried by the solar wind could be measured directly. When the wind blows past the Earth, it contains on average about five particles per cubic centimetre (mostly protons, the nuclei of hydrogen atoms) moving at about 500 km/sec (kilometres per second), but these numbers fluctuate greatly depending on the phase of the solar magnetic cycle and the presence or absence of recent flare activity.

Planets and their satellites.

With the exception of Mercury and Pluto, the orbits of the planets are all nearly circular; they lie within a few degrees of the same plane; and they have the same direct sense of revolution as the rotation of the Sun. Since these facts were first noted, they have suggested to philosophers and scientists such as Kant and Pierre-Simon Laplace of France that the planets of the solar system must have originally formed from a flat nebular disk that revolved about the primitive Sun. The exceptions, Mercury and Pluto, are not troublesome; they both suffer strong resonant interactions with other bodies that may have considerably modified their original orbital characteristics.

In the inner planetary system where the terrestrial planets--Mercury, Venus, Earth, and Mars--reside, the distance between successive planets is relatively small in comparison with the outer planetary system where the Jovian planets--Jupiter, Saturn, Uranus, and Neptune--reside. Moreover, the terrestrial planets are small and rocky or ironlike, while the Jovian planets (also called the giant planets) are large and gaseous or icy. Neither the terrestrial nor the Jovian planets exhibit the chemical elements in their cosmic proportions, but the latter, particularly Jupiter and Saturn, approach these proportions to a much closer degree. This implies that the process of planet building, unlike the mechanism of star formation, probably involves forces other than just gravity, for gravitation is universal and does not distinguish between different elements if they are in a gaseous form. Condensation (i.e., the separation of solid phases of matter from gaseous phases if the temperature drops to sufficiently low values) suggests itself as an important process.

From this point of view, the terrestrial planets have managed only to gather into their bodies mostly materials containing elements heavier than hydrogen and helium--materials such as silicate rocks and metallic iron or nickel, which can condense as solids from a gaseous phase even at relatively high temperatures (between 1,200 and 2,000 K). In contrast, Uranus and Neptune have not only accumulated rocky and metallic compounds but also ices of water, ammonia, and methane, which can condense from nebular gas only at much lower temperatures (between 100 and 200 K). Jupiter and Saturn succeeded additionally in capturing substantial amounts of hydrogen and helium (in their envelopes). Since hydrogen and helium at plausible nebular pressures do not solidify unless the temperature is lower than even in the coldest regions of interstellar space, this suggests that in the two largest planets of the solar system gravitation did play a role in the direct acquisition of massive amounts of these gases.

Pluto, which is small and icy and orbits farthest from the Sun, is not readily classifiable in the scheme outlined above. The discrepancy is not disruptive, however, because Pluto, discovered in 1930, and its moon, Charon, discovered in 1978, are relatively minor bodies similar in composition to the comets.

The terrestrial and Jovian planets possess other systematic differences: the former generally have no rings or satellites, while the latter each have a set of rings and many satellites. Here, Earth and Mars are exceptions to the rule. Earth has of course one satellite, the Moon; Mars has two, Phobos and Deimos. Of these exceptions, the more difficult case to explain has long remained the Moon because it is an unusually large object for a satellite. Indeed, the Moon is only somewhat smaller than the largest and most massive satellites in the solar system: Jupiter's Ganymede, Saturn's Titan, and Neptune's Triton. In comparison, Phobos and Deimos are tiny objects that may well have been captured after Mars had already formed.

The satellite and ring systems of the giant planets, particularly those of Jupiter and Saturn, resemble miniature planetary systems. As an analogy, one may say that moons and rings are to the giant planets what the planets and the asteroid belt are to the Sun. The moons of the giant planets can be classified as either regular or irregular. The regular satellites have nearly circular orbits lying in the same plane as the equator of the parent planet and revolve in the same direction as its rotation. The irregular satellites violate one or more of the above rules. In addition, they generally tend to be small bodies and to lie at large distances from the central planet. The regular satellites may have formed from protoplanetary disks that encircled the planet in the same manner as a protostellar disk encircled the Sun in the nebular hypothesis. The most likely explanation for the irregular satellites is that they are captured bodies.

The thin flat rings that encircle Jupiter, Saturn, Uranus, and Neptune are composed of innumerable small solid bodies. Each piece of the ring is in a nearly perfect circular orbit about the central planet. Theory suggests that noncircular motions are damped by mutual inelastic collisions of the particulate matter to very small values. These collisions would have led to gradual agglomeration into larger bodies had the rings not lain in such close proximity to the planet (i.e., within the Roche limit). The strong tidal forces that exist inside the Roche limit of a planet are believed to be capable of tearing apart loosely bound aggregates of particulate matter and thereby preventing their agglomeration into moons. It is unclear, however, whether planetary rings are the natural debris left over from an earlier period of satellite formation in a protoplanetary disk that extended almost to the planet's surface or whether they arose from the more recent breakup and erosion (by continual collisions and by micrometeoroid bombardment) of some larger parent body. There does exist some evidence from dynamic studies of the gravitational interactions of the rings and satellites of Saturn that the rings may be appreciably younger than the solar system in general.

Asteroids, meteoroids, comets, and interplanetary dust.

Asteroids are rocky or iron-bearing bodies found orbiting the Sun in great numbers in a belt between Mars and Jupiter. Nearly all of the total mass of the asteroids, about 10^-3 that of the Earth, is contained in the largest examples such as Ceres, Pallas, and Vesta, but the largest numbers have radii of one to 10 kilometres (the lower limit being more a matter of nomenclature than of measurement). A few bodies, as, for example, Chiron, lie outside the belt between Mars and Jupiter. The exceptions, however, are relatively rare. The theoretical understanding of this observational result lies in computer simulations that show that an asteroid placed almost anywhere else in the solar system besides the known asteroid belt would be unstable owing to gravitational perturbations by the planets. If the early solar system were littered with asteroid-sized bodies, then the emergence of the planets would have swept interplanetary space relatively clean except for the debris that happened to have orbits fit for survival.

Meteoroids are chunks of asteroids or comets that have Earth-crossing orbits. One theory for the production of meteoroids has them originating from the shattering of two asteroids that collide violently in space. Some of the pieces may subsequently suffer resonant interactions with Jupiter, which throw them in 10,000 to 100,000 years into elongated Earth-crossing orbits. A meteoroid entering the Earth's atmosphere will heat up during the passage and become a meteor, a fiery "shooting star." If the mass of the meteor exceeds one kilogram, it can survive the flight and land on the ground as a meteorite. Meteorites come in three basic compositions: stones, stony irons, and irons. Radioactive dating of meteorites establishes that they have a narrow range of ages. The time since their parent bodies first solidified equals about 4.6 10⁹ years, which yields the conventional estimate for the age of the entire solar system.

The cratering records on the airless (and therefore erosion-free) Moon and Mercury are consistent with a very heavy period of meteoritic impacts during the first several hundred million years of the history of the solar system, with the bombardment tailing off dramatically about 4 10⁹ years ago. This picture suggests that primitive asteroids and meteoroids may have been the building blocks ("planetesimals") of the terrestrial planets (and perhaps also the cores of the giant planets) and that the present-day asteroids failed to be gathered into another full-fledged planet because their noncircular velocities are so high (probably owing to the past near-resonant action of Jupiter's gravitational perturbations) as to cause them generally to shatter rather than to agglomerate when they collide.

Comets also are cosmic debris, probably planetesimals that originally resided in the vicinity of the orbits of Uranus and Neptune rather than in the warmer regions of the asteroid belt. Thus, the nuclei of comets are icy balls of frozen water, methane, and ammonia, mixed with small pieces of rock and dust, rather than the largely volatile-free stones and irons that typify asteroids. In the most popular theory, icy planetesimals in the primitive solar nebula that wandered close to Uranus or Neptune but not close enough to be captured by them were flung to great distances from the Sun, some to be lost from the solar system while others populated what was to become a great cloud of cometary bodies, perhaps 10 trillion in number. Such a cloud was first hypothesized by the Dutch astronomer Jan Hendrik Oort.

In the original version of the theory, the Oort cloud extended tens of thousands of times farther from the Sun than the Earth, a significant fraction of the way to the nearest stars. Random encounters with passing stars would periodically throw some of the comets into new orbits, plunging them back toward the heart of the solar system. As a comet nears the Sun, the ices begin to evaporate, loosening the trapped dust and forming a large coma that completely surrounds the small nucleus, which is the ultimate source of all the material. The solar wind blows back the evaporating gas into an ion tail, and radiation pressure pushes back the small particulate solids into a dust tail. Each solid particle is now an independently orbiting satellite of the Sun, and the accumulation of countless such passages by many comets contributes to the total quantity of dust particles and micrometeoroids found in interplanetary space. 

The total mass contained in all the comets is highly uncertain. Modern estimates range from 1 to 100 Earth masses. Part of the uncertainty concerns the reality of a hypothesized massive "inner Oort cloud" -- or "Kuiper belt" (if the distribution is flattened)--of comets that would exist at distances from the Sun 40 to 10,000 times that of the orbit of the Earth. At such locations, the comets would not be much perturbed by typical passing stars nor by the gravity of the planets of the solar system, and the comets could reside in the inner cloud or belt for long periods of time without detection. It has been speculated, however, that a rare close passage by another star (possibly an undetected companion of the Sun) may send a shower of such comets streaming toward the inner solar system. If enough large cometary nuclei in such showers happen to strike the Earth, the clouds of dust and ash that they would raise might be sufficient to trigger mass biological extinctions. An event of this kind appears especially promising for explaining the relatively sudden disappearance of the dinosaurs from the Earth. 

Origin of the solar system.

Because the chemical compositions of the planets as a function of increasing radial distance from the Sun follow a pattern that corresponds to sequential condensation from a gaseous state, cosmochemists originally postulated, for simplicity, that the solar nebula began in a hot and purely gaseous state. Small pieces of solids were then imagined to have condensed from the gas in the disk as the latter slowly cooled from high temperatures, with the coolest final temperatures being reached at the greatest distance from the centre. The process is akin to soot forming out of a smoking candle flame. Astronomical observations, however, show that dust grains of approximately the correct composition already exist in the interstellar medium, and theoretical calculations indicate that the refractory cores of the grains would survive introduction into most of the primitive solar nebula. The icy mantles that coat the grain cores would, however, be evaporated away in the inner solar system. It is probable, therefore, that the systematics of the observed planetary compositions reflect not a condensation sequence but rather an evaporation sequence.

In any case, whether the dust particles form by chemical condensation from the nebular gas or exist from the start, there seems little doubt that they would grow rapidly by various agglomeration processes and dissipatively settle into a thin layer of particulate matter in the midplane of the disk. Planetesimals of the sizes of asteroids and the nuclei of comets accumulate in this thin layer and further grow by gravitational processes into full-sized planets. The formation of the planets under these dissipative circumstances would explain why their orbits are nearly coplanar and circular.

Insofar as the planets first grow by the accumulation of solids, it is interesting to note that observations indicate all four Jovian planets to have rocky and icy cores containing 15-25 Earth masses. In addition to such cores, Jupiter and Saturn have hydrogen and helium envelopes amounting to about 300 and 70 Earth masses, respectively. This suggests, as theoretical calculations bear out, that 15-25 Earth masses represents a critical mass above which a growing planet in the solar nebula will begin to gravitationally gather nebular gas faster than it will accumulate solids. Indeed, once a protoplanet becomes massive enough, it can efficiently eject solid bodies as well as capture them. (The ones catapulted out by Jupiter and Saturn are likely to escape the system altogether.) In this way did Jupiter and Saturn become large and grow to occupy large areas.

Why Uranus and Neptune did not also gather massive gaseous envelopes is somewhat of a mystery. One possible theory is that, at the distances of Uranus and Neptune in the solar nebula, energetic radiation from the young Sun can dissociate hydrogen molecules and ionize the resultant atoms, heating the surface layers strongly enough (to about 10,000 K) to disperse the nebular gas over a period of about 10⁷ years. The full accumulation of the planetary cores of Uranus and Neptune probably took longer, and therefore their formation occurred in a relatively gas-free environment.

The growth of Pluto through the aggregation of many millions of cometlike bodies may have been limited by having to occur at the outermost fringes of the primitive solar nebula. Its moon, Charon, may have resulted either through fission of a rapidly rotating common parent body or through a late encounter and capture. Icy planetesimals that had close but noncolliding encounters with Uranus and Neptune either were thrown into the Sun (or into other planets) or now populate the Oort cloud of comets.

Interior to Jupiter the planets are all small. A plausible explanation follows from the observation that the solar nebula inside Jupiter's orbit may have been too hot to allow methane, ammonia, and water to exist in solid form. Computer simulations by the American geophysicist George Wetherill show that, restricted to the accumulation of only the rarer rocks and irons, the rapid runaway growth of planetesimals to embryos in the inner solar system stalls at masses comparable to the Moon's. Once a few hundred embryos of Moon-like masses have accumulated most of the solid matter in their immediate "feeding" zones, it takes them more than 10⁸ years gravitationally to pump up each other's eccentricities and aggregate through orbit crossings into four terrestrial planets.

A long duration for the formation of the terrestrial planets (supported by crater counts that indicate a prolonged period of bombardment extending over some 5 10⁸ years) suggests that Jupiter may have finished forming before the terrestrial planets did. A massive body at Jupiter's orbit may have then so stirred up the orbits of the planetesimals in the asteroid belt as to have prevented them from accumulating into a large body (see above). A fully formed Jupiter also may have stunted the growth of nearby Mars, explaining why Mars is so much smaller a terrestrial planet than either Venus or Earth.

The giant planets may also have sent fairly large bodies careening through the early solar system. In one version of the event, by the American astrophysicist Alastair G.W. Cameron and coworkers, a Mars-sized body crashed obliquely into the primitive Earth. The molten core of the intruder sank to the centre of the molten proto-Earth, but mantle material from both bodies went into orbit and eventually reaccreted into the Moon. The formation of the Moon from rocky substances would then explain why the lunar landings found the Moon to be much poorer in iron than the Earth.

A similar scenario purports to explain a compositional peculiarity in Mercury. A massive body from the asteroid belt sent close to the Sun would acquire such large velocities that on collision with Mercury it would splash off not only its own rocky mantle but much of Mercury's as well. An event of this kind might explain why Mercury has such a small rocky envelope in relation to its iron-nickel core when compared with the same features in Venus, Earth, and Mars.

Giant impacts would also add a chaotic element to the acquisition of planetary spins. Perhaps this accounts for the fact that, while most of the equators of the planets lie in roughly the same plane as their orbits about the Sun, Venus spins in a retrograde sense, whereas Uranus' spin axis is tilted over on its side. In reconstructing the details of the formation of the solar system, astronomers work under the handicap of not knowing whether certain special features arise as a general rule or as an exceptional circumstance.

Extrasolar planetary systems.

First, it is known from studies of gas clouds where stars are currently forming in the Galaxy that such regions generally rotate too quickly to collapse to a single normal star without any companions. Investigators know of many examples where the excess angular momentum has apparently been absorbed in the birth of a nearby orbiting star; indeed, binary stars are known to be the most common outcome of the star-formation process. It is, nevertheless, encouraging that infrared searches for faint companions around apparently single stars have found a few candidates for objects that lie intermediate to the least massive normal star and a giant planet such as Jupiter.

Second, infrared images taken from Earth-orbiting and ground-based telescopes have found flattened distributions of particulate solids encircling young stars that resemble the type of dusty nebular disk long hypothesized for the origin of the solar system. In a few cases, there have also been detections, from spectroscopic observations at millimetre and near-infrared wavelengths, of gaseous molecular material coextant with the solid particulates. These observations lend strong support to the view that the creation of planetary systems is likely to be a common by-product of the process of star formation.

STARS AND THE CHEMICAL ELEMENTS

The accepted interpretation of the abundance differences of Populations I and II is that stars synthesize heavy elements in their interiors. In the process of dying, some stars spew great quantities of this processed material into the gas clouds occupying the regions between the stars. The enriched matter then becomes incorporated into a new generation of forming stars, each successive generation having on average a greater proportion of heavy elements (and helium) than the last. During the 20th century astronomers have obtained considerable insight into why these processes should be the natural outcome of the structure and evolution of stars. (see also Index: stellar evolution)

STARS AND THE CHEMICAL ELEMENTS

The accepted interpretation of the abundance differences of Populations I and II is that stars synthesize heavy elements in their interiors. In the process of dying, some stars spew great quantities of this processed material into the gas clouds occupying the regions between the stars. The enriched matter then becomes incorporated into a new generation of forming stars, each successive generation having on average a greater proportion of heavy elements (and helium) than the last. During the 20th century astronomers have obtained considerable insight into why these processes should be the natural outcome of the structure and evolution of stars. 

Main-sequence structure of the stars.

Most of the time of the luminous stages of a star's life is spent on the main sequence, when it stably fuses hydrogen into helium in its core. The fusion process in a star with mass slightly greater than one solar mass is somewhat different from that in a star of one solar mass or less. In high-mass stars, hydrogen fusion occurs at high temperatures using preexisting nuclei of carbon and nitrogen as catalysts and, in the process, converting much of the carbon into nitrogen. In low-mass stars, hydrogen fusion occurs by direct combination of the hydrogen nuclei or their reaction products. The end product, however, is the same: the conversion of four hydrogen nuclei into one helium nucleus, with the release of the nuclear binding energy as a source of heat for the star. 

The time that a low-mass star spends on the main sequence differs drastically from that of a high-mass star. On the main sequence, a low-mass star spends its nuclear resource thriftily; a high-mass star, prodigiously. Hence, core hydrogen exhaustion for low-mass stars is delayed in comparison to high-mass stars. The main-sequence lifetime of a star half as massive as the Sun is about 3 10¹⁰ years, whereas that of a star of 50 solar masses would be roughly 3 10⁶ years.

Since the lifetime of a high-mass star is much less than the age of the Galaxy (roughly 10¹⁰ years) and since such stars exist during the present epoch in the Galaxy, the formation of high-mass stars must be an ongoing process. This is borne out by observations of the Galaxy and external galaxies, where bright blue stars are always found near giant clouds of gas and dust--the sites of both high-mass and low-mass star formation.

One of the most important tests for the theory of stellar structure and evolution comes from the examination of star clusters. Star clusters are gravitationally bound stellar groups that occur in two basic types: globular clusters, which typically are rich systems containing perhaps one million members distributed in a compact spherical volume with a strong concentration toward the centre, and open clusters, which typically are poor systems containing 1,000 members or fewer distributed loosely throughout an irregular volume. Globular cluster stars belong to Population II, while open cluster stars belong to Population I.

All astronomical observations of a star cluster indicate that its members formed from the same parent cloud. Thus, the stars in a cluster have the same age and the same initial compositions; the only notable difference among them is their masses. Since stars of different masses evolve at different rates, it should be possible to see a progression of evolutionary states as stars of increasing mass are considered. The effect is indeed seen, and the comparison of theoretical predictions with astronomical observations of star clusters yields one of the most satisfactory success stories of modern astrophysics. Such studies allow estimates of the ages of star clusters. The oldest turn out to be the globular clusters; they have ages estimated by various investigators between 1 10¹⁰ and 1.8 10¹⁰ years. Within the errors of the determinations, the ages of globulars are consistent with the expansion age of the universe--approximately 1.5 10¹⁰ years--obtained from Hubble's law. Thus, the globular cluster stars in the Galaxy must constitute some of the oldest stars in the Cosmos.

On the main sequence, a high-mass star is not only much more luminous than a low-mass star, but it also appears much bluer because its surface temperature is a few tens of thousands degrees instead of a few thousand degrees. The difference in surface temperature manifests itself not only in broadband colours but also in the pattern of atomic absorption lines that appear in spectroscopic diagnostics of the star. The Latin letters OBAFGKM are used to classify stars of different spectral types, with O stars having the hottest surface temperatures and M stars the coolest. The Sun is a G star. This classification scheme applies to all stars, not merely to those on the main sequence. To distinguish stars on the main sequence from those in different evolutionary states, astronomers introduced the concept of luminosity class. These categories are designated by Roman numerals from I to V, with I corresponding to supergiants and V to dwarfs. Main-sequence stars are dwarfs because stars have their smallest sizes as luminous objects when they shine by hydrogen fusion in the core, and a small star (dwarf or subgiant) of a given spectral type--i.e., surface temperature--radiates less than a large star (giant or bright giant or supergiant) of the same spectral type. Stars smaller than main-sequence stars are known (white dwarfs, neutron stars, or black holes), but they are very faint and are not normal stars and so are not assigned classifications in the normal scheme. 

About 90 percent of the luminous stars in a galaxy at any given time are on the main sequence. Most of the mass of a galaxy is contained in low-mass stars, but the small number of high-mass stars contributes a disproportionate fraction of the total light, especially at blue wavelengths. Most of the light at red wavelengths comes from evolved stars because all stars tend to become redder as they evolve from the main sequence (i.e., as their surfaces expand and cool). In addition, low-mass stars also tend to brighten as they age.

The end states of stars.

Astronomers believe that there are four possible end states for a star: (1) There may occur a violent explosion that completely overcomes self-gravity and disperses all constituent matter to interstellar space; this would leave nothing behind as the stellar remnant. (2) The free electrons in the core of the star may finally become so densely packed that quantum effects allow them to exert enough pressure (termed electron-degeneracy pressure; see below) to support the star even at zero temperature; this would leave behind a white dwarf as the stellar remnant. (3) If the mass of the core exceeds the maximum value -- the Chandrasekhar limit of 1.4 solar masses--allowed for a white dwarf, the compression of the stellar matter may finally be stopped at nuclear densities; this would leave behind a neutron star. (4) If the mass of the core is so large that even nuclear forces are incapable of supporting the star against its self-gravity, the gravitational collapse of the star may continue to a highly singular state at the centre; this would leave behind a black hole. 

Observations of star clusters and highly evolved objects suggest that stars initially less massive than about eight solar masses are able to lose enough of their envelopes in the final stages of normal stellar evolution that their burnt-out cores fall below the Chandrasekhar limit, resulting in a white dwarf remnant. Theoretical calculations are able to reproduce this result if empirical envelope-mass loss rates are adopted for the later stages of the evolution. In the range of 8 to 25 solar masses, the star is believed to suffer an iron-core collapse, giving an implosion of the central regions to form a neutron star and an expulsion of the envelope in a supernova explosion. Above 25 solar masses or so, the situation remains somewhat confused. Some stars may lose so much mass in powerful winds that their hydrogen envelopes are stripped clean. When they finally explode, they do so as supernovas of what astronomers term type Ib or Ic. In other stars, the energy deposited by neutrino emission (see below) may not suffice to blow off the outer layers, and the entire star collapses inward to form a black hole.

Observations of stellar remnants are reasonably in accord with the above picture. White dwarfs slowly cooling to the same temperature as the universe (3 K) seem to account for most of the dying stars, which is consistent with the fact that most stars are born with relatively low masses. At the sites of some historical supernova explosions, astronomers have found objects called pulsars, which are thought to be rotating magnetized neutron stars. And in some close binary systems, where a normal star is transferring matter to a compact companion, the companion can be inferred in different situations to be a white dwarf, a neutron star, or a black hole.

The evolution of stars.

The prediction that supernova explosions should liberate huge quantities of neutrinos found confirmation in the sudden brightening in 1987 of a previously known star in the Large Magellanic Cloud. The appearance of Supernova 1987A (SN 1987A), as this object was called, coincided with a burst of neutrino emission recorded by high-energy physics experiments originally designed to detect proton decay (see below). The magnitude and timing of the neutrino burst fit well with the model of the iron-core collapse of a star whose mass on the main sequence amounted to about 20 solar masses. Subsequent measurements of the light curve demonstrated that, in general agreement with nucleosynthetic expectations, SN 1987A ejected about 0.07 solar mass of the radioactive isotope nickel-56, with a half-life of 6 days, which decays into cobalt-56, with a 77-day half-life, and then into stable iron-56.

Another interesting by-product of the supernova mechanism described above is that large numbers of free neutrons can be liberated in the envelope. Seed nuclei can capture these free neutrons to become heavier and eventually create many of the elements beyond iron in the periodic table, including radioactive species like uranium. Different isotopes of uranium decay at different rates, and knowing the primitive ratios in which supernovas create these isotopes enables radiochemists to compute, from the corresponding measured values in uranium ore, the elapsed time since these isotopes were produced and introduced into the solar system. Depending on the rates of supernova explosions in the history of the Galaxy, these calculations indicate that uranium synthesis began between 6 10⁹ and 1.5 10¹⁰ years ago. This, then, is another method for independently estimating the age of the Galaxy. Again, within the uncertainties of the determination, the value is consistent with the Hubble expansion age 

The fundamental difference in evolutionary outcomes between high-mass and low-mass stars can be traced to the theory of white dwarfs. Basically, every star eventually tries to generate a white dwarf at its core as it evolves and undergoes core contraction. During the 1920s, with the dawn of modern quantum mechanics, the British physicist Ralph H. Fowler showed that a white dwarf has the peculiar property that the more massive it is, the smaller its radius. The reason is relatively simple: a more massive white dwarf has more self-gravity, and so more pressure is required to counter the stronger gravity. Pressure increases when the degenerate electron gas constituting a white dwarf is compressed; it becomes strong enough to balance gravitational force only at very great densities. Consequently, equilibrium between the internal degeneracy pressure and the force of gravity is reached at a smaller size for a more massive white dwarf. 

The American astrophysicist Subrahmanyan Chandrasekhar made a crucial modification to this hypothesis in order to accommodate Einstein's special theory of relativity. Chandrasekhar showed that relativistic effects imposed an upper limit on the mass of possible white dwarfs. This limit arises because electrons cannot move faster than the speed of light; there comes a point where the increase in internal degeneracy pressure is no longer able to keep the self-gravity from literally trying to crush the star to zero size. For likely white-dwarf compositions, this limit corresponds to 1.4 solar masses as noted above. 

Consider a star that attempts to exceed the Chandrasekhar limit, assuming that it has enough material--even after envelope-mass loss--to try to build a massive white dwarf by depositing layer after layer of nuclear ash into its core. As the limit is approached, the core's outer boundary shrinks almost to arbitrarily small dimensions, generating above it enormous gravitational fields. To counteract the gravity, the pressures in the shell above the core must rise correspondingly, yielding densities and temperatures that are as high as needed to drive all thermonuclear reactions to completion. If nothing else intervenes, this situation must end in the iron catastrophe described above.

In contrast, in a low-mass star the final mass of the core may end up well below the Chandrasekhar limit. The shells outside the core may still become dense and hot enough to yield copious amounts of hydrogen and helium fusion, and this heat input into the envelope will greatly distend the envelope of the star, bringing the star to the red giant and red supergiant evolutionary phases that characterize the later stages. The outer atmospheres of such stars are often cool enough to allow the condensation of some of the heavy elements into solid particles. Dust grains composed of a rocky silicate are probably the most common outcome, but graphite or silicon carbide grains are possibilities in carbon-rich stars. In any case, because the envelope of the star is so extended, the surface gravity is too weak to hold the atmospheric mix of gas and dust, and this mixture blows out of the star as a prodigious stellar wind. Objects in this state are called planetary nebulas. The observed loss of matter occurs at a rate rapid enough to strip off the entire envelope, revealing eventually a white-hot core that is now a bare white dwarf. Since the mass loss reduced the stellar mass below the Chandrasekhar limit, the core never progressed to very advanced stages of nuclear fusion, giving the most common white dwarfs in the Galaxy a likely composition of carbon and oxygen. 

Interstellar clouds.

Atomic hydrogen clouds are the most widely distributed in interstellar space and, together with molecular hydrogen clouds, contain most of the gaseous and particulate matter of interstellar space. Molecular hydrogen clouds contain a wide range of molecules besides the hydrogen molecule H₂ and for that reason are simply called molecular clouds. Ionized hydrogen clouds, called H II regions by astronomers, are fluorescent masses of gas, such as the famous Orion Nebula, which have been lit up by hot blue stars recently born from the neutral gas, the hydrogen becoming dissociated and ionized because of the copious outpouring of ultraviolet photons from such massive stars.

Dust particles are suspended in all three types of clouds, and their effects can be seen in the absorption and scattering of optical light or in the thermal emission of infrared radiation. The refractory cores of the dust grains were probably expelled from the atmospheres of countless red giant stars, although icy mantles may be acquired in molecular clouds by the adhesion of molecules to the cold grain surfaces when they collide. It has been estimated that dust grains typically account for 1 percent of the mass of an interstellar cloud. Because the internal constitution of dust is primarily elements heavier than hydrogen and helium and because the cosmic mass fraction of all such elements is only a few percent of the total, dust grains must contain a significant fraction of the total cosmic abundance of heavy elements. This deduction is in accord with the observational finding that many heavy elements are severely underrepresented in the gas phase of interstellar clouds. They presumably have condensed out as solid particles. 

Of greatest interest to the present discussion are the molecular clouds, because it is from giant complexes of such clouds that most stars are formed. Radiative cooling by the molecules and dust in them keeps the matter at very low average temperatures, about 10 K, and at relatively high densities as compared with atomic hydrogen clouds. These two circumstances, combined with the large mass (10⁵ or 10⁶ solar masses) of a typical giant molecular cloud complex, make molecular clouds ideal sites for star formation because, even with dimensions spanning hundreds of light-years, they are held together by their self-gravitation. Once a gaseous astronomical body becomes self-gravitating, the formation of still more condensed states--in this case, stars--is almost inevitable. 

Star formation.

Low-mass stars also are formed in associations called T associations after the prototypical stars found in such groups, T Tauri stars. The stars of a T association form from loose aggregates of small molecular cloud cores a few tenths of a light-year in size that are randomly distributed through a larger region of lower average density. The formation of stars in associations is the most common outcome; bound clusters account for only about 1 to 10 percent of all star births. The overall efficiency of star formation in associations is quite small. Typically less than 1 percent of the mass of a molecular cloud becomes stars in one crossing time of the molecular cloud (about 5 10⁶ years). Low efficiency of star formation presumably explains why any interstellar gas remains in the Galaxy after 10¹⁰ years of evolution. Star formation at the present time must be a mere trickle of the torrent that occurred when the Galaxy was young.

A typical cloud core rotates fairly slowly, and its distribution of mass is strongly concentrated toward the centre. The slow rotation rate is probably attributable to the braking action of magnetic fields that thread through the core and its envelope. This magnetic braking forces the core to rotate at nearly the same angular speed as the envelope as long as the core does not go into dynamic collapse. Such braking is an important process because it assures a source of matter of relatively low angular momentum (by the standards of the interstellar medium) for the formation of stars and planetary systems. It also has been proposed that magnetic fields play an important role in the very separation of the cores from their envelopes. The proposal involves the slippage of the neutral component of a lightly ionized gas under the action of the self-gravity of the matter past the charged particles suspended in a background magnetic field. This slow slippage would provide the theoretical explanation for the observed low overall efficiency of star formation in molecular clouds.

At some point in the course of the evolution of a molecular cloud, one or more of its cores become unstable and subject to gravitational collapse. Good arguments exist that the central regions should collapse first, producing a condensed protostar whose contraction is halted by the large buildup of thermal pressure when radiation can no longer escape from the interior to keep the (now opaque) body relatively cool. The protostar, which initially has a mass not much larger than Jupiter, continues to grow by accretion as more and more overlying material falls on top of it. The infall shock, at the surfaces of the protostar and the swirling nebular disk surrounding it, arrests the inflow, creating an intense radiation field that tries to work its way out of the infalling envelope of gas and dust. The photons, having optical wavelengths, are degraded into longer wavelengths by dust absorption and reemission, so that the protostar is apparent to a distant observer only as an infrared object. Provided that proper account is taken of the effects of rotation and magnetic field, this theoretical picture correlates with the radiative spectra emitted by many candidate protostars discovered near the centres of molecular cloud cores.

An interesting speculation concerning the mechanism that ends the infall phase exists: it notes that the inflow process cannot run to completion. Since molecular clouds as a whole contain much more mass than what goes into each generation of stars, the depletion of the available raw material is not what stops the accretion flow. A rather different picture is revealed by observations at radio, optical, and X-ray wavelengths. All newly born stars are highly active, blowing powerful winds that clear the surrounding regions of the infalling gas and dust. It is apparently this wind that reverses the accretion flow.

The geometric form taken by the outflow is intriguing. Jets of matter seem to squirt in opposite directions along the rotational poles of the star (or disk) that sweep up the ambient matter in two lobes of outwardly moving molecular gas--the so-called bipolar flows. Such jets and bipolar flows are doubly interesting because their counterparts were discovered some time earlier on a fantastically larger scale in the double-lobed forms of extragalactic radio sources 

The underlying energy source that drives the outflow is unknown. Promising mechanisms invoke tapping the rotational energy stored in either the newly formed star or the inner parts of its nebular disk. There exist theories suggesting that strong magnetic fields coupled with rapid rotation act as whirling rotary blades to fling out the nearby gas. Eventual collimation of the outflow toward the rotation axes appears to be a generic feature of many proposed models.

Pre-main-sequence stars of low mass first appear as visible objects, T Tauri stars, with sizes that are several times their ultimate main-sequence sizes. They subsequently contract on a time scale of tens of millions of years, the main source of radiant energy in this phase being the release of gravitational energy. When their central temperatures reach values comparable to 10⁷ K, hydrogen fusion ignites in their cores, and they settle down to long stable lives on the main sequence. The early evolution of high-mass stars is similar; the only difference is that their faster overall evolution may allow them to reach the main sequence while they are still enshrouded in the cocoon of gas and dust from which they formed. 

GALAXIES

The Milky Way Galaxy.

A hint of the processes of the formation and evolution of the Galaxy is contained in the general correlation between the spatial location of a star in the galactic system and its heavy-element abundance. The stars found in the disk of the Galaxy are mostly Population I stars; those in the halo are of the Population II type; and those in the bulge are a mixture of the two. This correlation was first noticed in the 1940s by the American astronomer Walter Baade from his investigation of the Andromeda galaxy. Since the theory of nucleosynthesis states that the abundance of heavy elements in successive generations of stars should increase with age, it can be deduced that star formation in the halo terminated long ago, while it has continued in the disk to the present day.

The shapes acquired by the different stellar components can be understood in terms of the orbital characteristics of the different stellar populations. For Population I stars, the motion corresponds nearly to circular orbits in a single plane; the random velocities above the circular component are small, accounting for the flattened shape of the galactic disk. For Population II stars, the noncircular velocities are much larger; the stars orbit randomly about the Galaxy like a swarm of bees around a hive, accounting for the spheroidal shapes of the galactic bulge and halo.

In 1962 Olin Eggen of Australia, Donald Lynden-Bell of England, and Allan Sandage of the United States pieced together the chemical and kinematic lines of evidence to argue that the Galaxy must have originated through the coherent dynamic collapse of a single large gas cloud, in which the stars of the halo condensed quickly (within about 2 10⁸ years) from the gas, to be followed by the formation of the bulge and disk. Subsequent discoveries that the globular clusters of the halo have a spread of heavy-element abundances and probable ages and that some stars in the bulge are as old or older than the oldest stars in the halo have cast doubt on this simple view. An alternative scenario pictures the Galaxy to have built up relatively slowly over a period of a few times 10⁹ years through the agglomeration of smaller galactic fragments. Some astronomers believe that a "thick-disk" component reported for the Milky Way system and other galaxies arise by this process, but too great a thickening of the layer of stars in the disk may result if the captured companions have more than about 10 percent of the Galaxy's mass.

Although the velocities of the stars within a few thousand light-years of the Sun in the direction perpendicular to the galactic plane are generally small, they are not zero. By investigating the statistics of these motions and the vertical structure of the disk, it is possible to deduce the vertical component of the gravitational field of the Galaxy and thereby the total mass of material required locally to supply the observed gravity. The quantity of required material is called Oort's limit (after the aforementioned J.H. Oort), and it exceeds by a factor of about two the quantity of available material, as observed in the form of known stars and gas clouds. This result constitutes the closest example of a general discrepancy arising on galactic scales whenever dynamically derived masses are compared with direct counts of observationally accessible objects. The missing matter in Oort's limit refers, however, to a flattened population and may differ in ultimate resolution from the more general dark-matter problem (see below), which is associated with the halos of galaxies and beyond.

From star counts, one can derive another quantity of astronomical interest, the mean brightness (per unit area) in the solar neighbourhood. If one divides this quantity into the mass (per unit area) corresponding to Oort's limit, one obtains the local mass-to-light ratio, which astronomers have measured to be about five in solar units. In other words, the gravitating mass in the Galaxy has a mean efficiency for producing light that is five times less than the Sun's. This implies, first, that the average star must be less massive than the Sun and, second, that the amount of helium presently inside stars--in contrast with the heavier elements--cannot have been produced by stellar processes. The reason is simple. The Sun, with a mass-to-light ratio of unity, will manage to convert about 10 percent of its mass (in the core) into helium in 10¹⁰ years (after which it leaves the main sequence); matter with a mean mass-to-light ratio of five, therefore, would convert only 2 percent of its mass to helium in 10¹⁰ years, roughly the age of both the Galaxy and the universe. The cosmic abundance of helium is approximately 26 or 27 percent of the total mass; thus, unless the Galaxy was much brighter in the past than it is today (for which there is no observational evidence), the bulk of the helium in the universe must have been created by nonstellar processes. Astronomers now believe that a primordial abundance of helium of about 24 percent by mass emerged from the big bang. Among other arguments, this is the value derived from the analyses of the chemical compositions of H II regions in external galaxies where the heavy-element abundance is very low and where, therefore, nuclear processing by stars has presumably been small. 

It is possible, of course, to examine the statistics of the random velocities of stars in the two directions parallel to the galactic plane as well as in the vertical direction. The Swedish astronomer Bertil Lindblad was the first to carry out such an analysis. His work, combined with Oort's study in 1927 of the constants of the differential rotation of the Galaxy, gave the period of revolution of stars such as the Sun about the galactic centre. The modern value for this period equals about 2.5 10⁸ years. With Shapley's measurement of the distance to the galactic centre and with the assumption that stars like the Sun circle the Galaxy because they are gravitationally bound to it, it is possible to estimate the total mass interior to the solar distance from the galactic centre. Modern estimates yield roughly 2 10¹¹ solar masses. Since the Sun is somewhat more massive than the typical star, the Galaxy must contain more than 10¹¹ stars.

Detailed information can be gleaned about the distribution of mass in the Galaxy if one possesses a knowledge of the rotational speeds of disk matter at other radial locations in the Galaxy. The most common measurements are of atomic hydrogen in its spin-flip transition at 21-centimetre wavelength and of the carbon monoxide molecule in one or another of its rotational transitions at millimetre wavelengths. These observations also provide data concerning the total amount of atomic and molecular hydrogen gas contained in the Galaxy. To convert the carbon monoxide abundance to a molecular hydrogen abundance (which cannot be measured directly except at ultraviolet wavelengths that suffer tremendous dust extinction) requires a complicated series of calibrations of nearby sources. The mass of gas in the Galaxy is a few times 10⁹ solar masses, about evenly divided between atomic and molecular hydrogen clouds. Most of the observed mass of the Galaxy is in the form of stars; gas and dust make up only a few percent of the total.

By a combination of such measurements, astronomers can obtain the rotation curve of the Galaxy from its innermost regions to a radial distance of almost 60,000 light-years from the galactic centre. This rotation curve implies that the mass of the Galaxy measured out to a certain distance r does not converge to a fixed value as r increases but continues to rise roughly in linear proportion to r. The mass contained interior to the most distant radius measured amounts to about 5 10¹¹ solar masses. Observations indicate, however, that the integrated light from a galaxy like the Milky Way system does not increase similarly with increasing r but approaches asymptotically a finite value. Thus, the local mass-to-light ratio of the Galaxy, like those of other spiral galaxies, must increase dramatically toward its outer parts where the halo dominates. Another way to state the problem is that the observed rotational velocities of gas clouds in the outer parts of spiral galaxies are so large that they would not be bound to the galaxies unless the galaxies were more massive than inferred from direct measurements of their stellar and gas contents. Most astronomers now accept the likelihood of dark halos that contain as much mass as is present in the visible disks and bulges; more controversial are the claims that these halos may increase known galactic masses by factors of 10 or 100. 

Classification of galaxies.

Ellipticals and spirals constitute the two largest reservoirs of the stars in the universe, and the placement of individual galaxies into these two major categories is refined by adding a numeral 1 through 7 or a letter "a" through "c" to their designation. The sequence E0 to E7 denotes one of increasing flattening (as seen in projection in the sky). The sequence Sa to Sc, or SBa to SBc, represents decreasing tightness of winding of the spiral arms and decreasing size of the central bulge relative to the disk.

A useful analogy with stars is the introduction by Sidney van den Bergh of Canada of the concept of luminosity class. The scheme appends to the Hubble type a luminosity-class label, from Roman numeral I for the intrinsically brightest (and most massive) spiral galaxies to Roman numeral V for the intrinsically faintest (and least massive) spirals. The utility of this scheme, as applied to spirals, rests with the fact that it is possible to assign them a luminosity class without actually measuring their distance (to obtain an absolute brightness from an observed apparent brightness). The luminosity class of a spiral galaxy correlates well with the regularity (or "prettiness") of the spiral structure: in class I galaxies the arms are long and well developed and have a high surface brightness; in class III they are patchy and fuzzy; and in class V there may be barely a hint of a spiral structure. Elliptical galaxies, lacking spiral arms, cannot have their absolute brightnesses estimated by the same morphological considerations; hence, the concept of a luminosity class for them is less empirically useful. When the masses of elliptical galaxies at known distances are deduced from measured velocities or apparent luminosities, they range from a few million solar masses (dwarf ellipticals) to more than 10¹² solar masses (giant ellipticals). Thus, giant ellipticals and giant spirals have comparable masses. Yet, it should be noted that the very largest elliptical galaxies in the universe, the supergiant cD systems, are unique and perhaps have masses approaching 10¹⁴ solar masses in some extreme cases.  

Dynamics of ellipticals and spirals.

The dominant motion in the disks of normal spiral galaxies is differential galactic rotation, with the random motions of stars being relatively small and that of the atomic and molecular gas smaller still. A surprising result is that the rotation curves of almost all well-studied spiral galaxies become flat at large radial distances. As one goes out from the centre, the rotational velocity rises to a constant value V and then maintains it for as far as one can make the measurements. This implies, as already noted for the Milky Way Galaxy, that the mass contained within r increases linearly with increasing r and provides the firmest piece of evidence in support of the hypothesis that large amounts of dark matter may be present in the halos of spiral galaxies. 

The qualitative fact of disk galaxies rotating differentially, with the inner parts having shorter rotational periods than the outer parts, has been known since Lindblad's and Oort's investigations of the problem for the Milky Way system in the 1920s. This fact, combined with age estimates for all galaxies of about 10¹⁰ years, presents a dilemma for the origin of spiral structure. If spiral arms are viewed as consisting always of the same material (e.g., the same gas clouds that give birth to the brilliant OB stars and H II regions that best define the optical spiral structure), then the arms should wind up. In particular, with a flat rotation curve, material at half the solar distance from the galactic centre should go around twice for each revolution of the material at the solar distance, and an extra turn should then be added to each spiral arm between these two radii every 2.5 10⁸ years. This would give the spiral arms of the Galaxy (and other spiral galaxies like it) several dozens of turns over the lifetime of the Galaxy, whereas spiral galaxies have in fact never been observed with more than one or two turns.

A way out of the winding dilemma is the proposal that spiral structure is a wave phenomenon, the spiral arms being a local "piling-up" of stars and gas clouds that individually flow through the spiral pattern, much as a traffic jam is a local piling-up of cars and trucks that individually flow through the jam. The piling-up arises because the self-gravity of the excess matter in the arms causes deflections of what would otherwise be circular orbits (on average), the deflections self-consistently producing the original pileup. Most astronomers are agreed that density waves underlie the phenomenon of spiral structure in the so-called grand-design galaxies. More controversial is whether some other mechanism (e.g., "stochastic star formation") might play a role in galaxies where the spiral structure is "flocculent."

In modern density-wave theory, as developed by the American mathematician Chia-chiao Lin and his associates, spiral structure represents an unstable mode of collective oscillation. The instability provides a way by which a differentially rotating disk galaxy may release free energy of differential rotation and spontaneously generate spiral waves. The balance of the growth of these waves against their dissipation (through the response of the interstellar gas clouds) may yield a quasi-stationary state whereby gaseous matter slowly drifts to the interior and angular momentum is steadily transported to the exterior. Although the details of the entire picture remain incomplete, many of the basic predictions--as, for example, that the perturbations in density and velocity should be strongest in the component with the smallest random velocities (i.e., gas and dust clouds)--have already been confirmed both qualitatively and quantitatively in several well-observed spiral galaxies. Furthermore, the Hubble correlation between the tightness of spiral windings and the size of the central bulge relative to the disk, as well as the van den Bergh correlation between luminosity class and the degree of organization of the spiral structure, are simple direct consequences of density-wave theory.

A similar explanation probably underlies the barred spiral galaxies, with the basic underlying disturbance being an oval distortion. The predicted departures from circular motions are larger in barred spirals than in ordinary spirals, and this seems to be consistent with the observational evidence that currently exists for this problem. The enhanced rates at which matter is brought to the centres of such galaxies may have implications for various energetic events that take place in some galactic nuclei. It has even been proposed on the basis of observed peculiarities of gas motions and various infrared images that the central regions of the Milky Way Galaxy may contain a small bar.

In elliptical galaxies, the constituent stars have random velocities that are generally much larger than the rotational motions. This explains why ellipticals possess neither thin disks nor spiral arms. Moreover, giant ellipticals are flatter than would be inferred from the amount of rotation that they do possess, and increasing rotation does not necessarily lead to increasing flattening, as appears to happen, for example, to ellipticals of lower luminosity and the bulges of spiral galaxies. Also, most ellipticals do not appear to have young stars, probably because the small measurable amounts of gas and dust that exist in them cannot support an active rate of star formation.

Mathematical analysis and computer simulations since the early 1970s suggest a possible stellar-dynamic basis for understanding the basic shapes of giant elliptical galaxies. Unlike the bulges and disks of S0 galaxies, the bodies of giant ellipticals may not be figures of revolution (e.g., oblate spheroids) but may possess three axes of unequal lengths. In the models, the triaxial shape arises because the random velocities of the stars are anisotropic (not equal in all directions). Such a state of affairs seems consistent with the existing observational data, in particular the finding in several ellipticals that significant rotation exists around the longest apparent axis. A healthy fraction of nearby ellipticals, moreover, show rapidly rotating cores, which may represent the remains of captured dwarf galaxies that have spiraled to the centres of their larger hosts.

An interesting empirical property shared by both ellipticals and spirals is that their luminosities L seem to be proportional to the fourth power of their random or circular velocities V. The proportionality constant can be calibrated with the help of nearby (giant) galaxies, and the resulting relation may then be used for cosmological investigations. In particular, the determination of distances is a recurring astronomical problem, and the relation, L proportional to V⁴, provides a method for obtaining distances. In brief, a measurement of V allows the determination of L, which, combined with the observed apparent brightness, gives the distance of the object. 

Interacting galaxies.

Stripping (of matter), merging (of the main bodies of the galaxies), and sinking (of the satellite galaxy toward the centre of the host) are all represented in the above example, and these processes, individually and collectively, have been invoked by theorists in a wide variety of contexts and by a wide variety of names to explain different observed galactic phenomena. The most interesting application is perhaps to the origin of elliptical galaxies.

It has been proposed by the American astronomer-mathematician Alar Toomre that elliptical galaxies result from the merger of spiral galaxies, jumbled piles of stars from the wreckage of collisions of bound pairs of galaxies with arbitrarily oriented spins and orbits. A potential difficulty with the original theory was the fate of the interstellar gas and dust. Considerable evidence has since accumulated (i.e., with the launch in 1983 of the Infrared Astronomical Satellite [IRAS]) to show that tidal interactions and galactic mergers can induce strong bursts of star formation that use up the interstellar material at rates up to 100 times faster than in normal galaxies (see below). An extension of similar ideas suggests that the supergiant ellipticals, the cD galaxies that tend to lie at or near the centres (or density maxima) of rich clusters of galaxies, grew bloated by "cannibalizing" their smaller neighbours. 

Galaxy formation.

Material that reached a completely flattened state while still in a gaseous state would have its vertical component of motion arrested in a strong shock wave and form the disk of a galaxy. Material that formed dense stars or protostars on the way down would be able to pass through the disk virtually unimpeded and, after several bounces, would settle to form the bulge and halo of a disk galaxy like the Milky Way system. It was also thought that a slight modification could produce elliptical galaxies--namely, if the efficiency of star formation were so high during the collapse phase that virtually all the matter turned into stars before flattening into a disk, then a single quasi-spherical stellar component might result. Given the developments since the 1970s described above, however, serious doubts have been raised against this scenario. The spread in ages and heavy-element abundances of halo and bulge stars in the Milky Way Galaxy, the anisotropic distribution of stellar velocities in elliptical galaxies, and the statistics of starburst galaxies and interacting galaxies all argue for the importance of galactic mergers (perhaps involving predominantly dwarf systems) in the buildup of giant galaxies. 

There also exists observational evidence that galaxies existent at a time corresponding to a redshift of three or four have properties quite different from those that exist today at redshifts near zero (see Cosmological models below). High-redshift galaxies can be found in association with strong extragalactic radio sources (see below Quasars and related objects), and, when such galaxies are imaged optically, they often show complex lumpy structures suggestive of recent mergers and interactions. A similar result applies when distant galaxies were imaged in a random fashion by the Hubble Space Telescope, the Earth-orbiting observational system launched in 1990.

It remains uncertain, however, when the first stars in any galactic-sized lump formed. Infrared studies demonstrate that well-developed stellar populations already exist in galaxies with redshifts of a few and perhaps even 5 or 10. The observational discovery of a genuine primeval galaxy would remove many uncertainties. In a collapse environment involving only hydrogen and helium gas, the primary diagnostic would be the copious emission of Lyman-alpha radiation (corresponding to the transition between the first excited state and the ground state of atomic hydrogen). The rest wavelength of this transition lies in the ultraviolet, but in primeval galaxies the cosmological redshift would make the observed wavelength longer (i.e., toward the red end of the spectrum). From this point of view, it is interesting that searches near known quasars have uncovered Lyman-alpha-emitting galaxies with redshifts exceeding three. 

There exists a body of opinion that the stars of a primeval galaxy will generate dust at such a rapid rate that all intrinsic Lyman-alpha production by the galaxy will be degraded to thermal infrared-continuum radiation. In this case, primeval galaxies may resemble the "starburst galaxies" that were discovered by IRAS. In contrast to normal galaxies like the Milky Way system where the ratio of infrared to visible luminosities is about unity, these sources can emit up to 100 times more infrared radiation than visible light. The only viable explanation for the infrared excess is that these galaxies are somehow undergoing enormous bursts of star formation. Ground-based observations that followed up the IRAS discovery showed that the activity in starburst galaxies is often confined to the central portions of the systems and that many of the candidate sources correspond either to interacting galaxies or to barred spirals. This suggests that starbursts may be triggered by the gravitational perturbations that have brought large amounts of molecular gas to the central regions of the galaxy. A similar burst of star formation might be expected to occur in an era when the matter of a galaxy was nearly all gas rather than all stars. Since astronomers have not found any general evidence for such large-scale energetic events, it becomes plausible to contemplate the formation of giant galaxies as a more protracted process (through the mergers of many dwarf systems), extending possibly even to the present epoch. 

QUASARS AND RELATED OBJECTS

Galaxies are where astronomers find stars, the major transformers of matter into energy in the universe. Paradoxically, it is also from the study of galaxies that astronomers first learned that there exist in the universe sources of energy individually much more powerful than stars. These sources are radio galaxies and quasars, and their discovery in the 1950s and '60s led to the establishment of a new branch of astronomy, high-energy astrophysics.

Extragalactic radio sources.

The most powerful extragalactic sources of radio waves are double-lobed sources (or "dumbbells") in which two large regions of radio emission are situated in a line on diametrically opposite sides of an optical galaxy. The parent galaxy is usually a giant elliptical, sometimes with evidence of recent interaction. The classic example is Cygnus A, the strongest radio source in the direction of the constellation Cygnus. Cygnus A was once thought to be two galaxies of comparable size in collision, but more recent ideas suggest that it is a giant elliptical whose body is bifurcated by a dust lane from a spiral galaxy that it recently swallowed. The collisional hypothesis in its original form was abandoned because of the enormous energies found to be needed to explain the radio emission.

The radio waves coming from double-lobed sources are undoubtedly synchrotron radiation, produced when relativistic electrons (those traveling at nearly the speed of light) emit a quasi-continuous spectrum as they gyrate wildly in magnetic fields. The typical spectrum of the observed radio waves decreases as a power of increasing frequency, which is conventionally interpreted, by analogy with the situation known to hold for the Galaxy in terms of radiation by cosmic-ray electrons, with a decreasing power-law distribution of energies. The radio waves typically also show high degrees of linear polarization, another characteristic of synchrotron radiation in well-ordered magnetic fields.

A given amount of received synchrotron radiation can be explained in principle by a variety of assumed conditions. For example, a high energy content in particles (relativistic electrons) combined with a low content in magnetic fields will give the same radio luminosity as a low energy content in particles combined with a high content in magnetic fields. The American astrophysicist Geoffrey R. Burbidge showed that a minimum value for the sum results if one assumes that the energy contents of particles and fields are comparable. The minimum total energy computed in this way for Cygnus A (whose distance could be estimated from the optical properties of the parent galaxy) proved to be between 10⁶⁰ and 10⁶¹ ergs.

Schmidt's discovery raised immediate excitement, since 3C 273 had a redshift whose magnitude had been seen theretofore only among the most distant galaxies. Yet it had a starlike appearance, with an apparent brightness (but not a spectrum) in visible light not very different from that of a galactic star at a distance of a few thousand light-years. If the quasar lay at a distance appropriate to distant galaxies a few times 10⁹ light-years away, then the quasar must be 10¹² times brighter than an ordinary star. Similar conclusions were reached for other examples. Quasars seemed to be intrinsically brighter than even the most luminous galaxies known, yet they presented the pointlike image of a star.

A hint of the actual physical dimensions of quasars came when sizable variations of total light output were seen from some quasars over a year or two. These variations implied that the dimensions of the regions emitting optical light in quasars must not exceed a light-year or two, since coherent fluctuations cannot be established in any physical object in less time than it takes photons, which move at the fastest possible speed, to travel across the object. These conclusions were reinforced by later satellite measurements that showed that many quasars had even more X-ray emission than optical emission, and the total X-ray intensity could vary in a period of hours. In other words, quasars released energy at a rate exceeding 10¹² suns, yet the central machine occupied a region only the size of the solar system.

Understandably, the implications were too fantastic for many people to accept, and a number of alternative interpretations were attempted. An idea common to several of the alternatives involved the proposal that the redshift of quasars arose from a different (i.e., noncosmological) origin than that accepted for galaxies. In that case, the distance to the quasars could be much less than assumed to estimate the energy outputs, and the requirements might be drastically relaxed. None of the alternative proposals, however, withstood close examination.

In any case, there now exists ample evidence for the validity of attributing cosmological distances to quasars. The strongest arguments are the following. When the strong nonstellar light from the central quasar is eliminated by mechanical or electronic means, a fuzzy haze can sometimes be detected still surrounding the quasar. When this light is examined carefully, it turns out to have the colour and spectral characteristics appropriate to a normal giant galaxy. This suggests that the quasar phenomenon is related to nuclear activity in an otherwise normal galaxy. In support of this view is the observation that quasars do not really form a unique class of objects. For example, not only are there elliptical galaxies that have radio-emission characteristics similar to those of quasars, but there are weaker radio sources among spiral galaxies (called Seyferts after their discoverer, the American astronomer Carl K. Seyfert), which have bright nuclei that exhibit qualitatively the same kinds of optical emission lines and nonstellar continuum light seen in quasars. There also are elliptical galaxies, N galaxies, and the so-called BL Lac objects, which have nuclei that are exceptionally bright in optical light. Plausible "unification schemes" have been proposed to explain many of these objects as the same intrinsic structure but viewed at different orientations with respect to relativistically beamed jets or with obscuring dust tori surrounding the nuclear regions or both. Finally, a number of quasars--including the closest example, the famous source 3C 273--have been found to lie among clusters of galaxies. When the redshifts of the cluster galaxies are measured, they have redshifts that bracket the quasar's, suggesting that the quasar is located in a galaxy that is itself a cluster member. 

Black-hole model for active galactic nuclei.

The black hole has to be supermassive for its gravitational attraction to overwhelm the strong radiation forces that attempt to push the accreting matter back out. For a luminosity of 10⁴⁶ erg/sec, which is a typical inferred X-ray value for quasars, the black hole must exceed 10⁸ solar masses. The event horizon of a 10⁸ solar-mass black hole, from inside which even photons would not be able to escape, has a circumference of about two light-hours. Matter orbiting in a circle somewhat outside of the event horizon would be hot enough to emit X rays and have an orbital period of several hours; if this material is lumpy or has a nonaxisymmetric distribution as it disappears into the event horizon, variations of the X-ray output on a time scale of a few hours might naturally be expected.

To produce 10⁴⁶ erg/sec, the black hole has to swallow about two solar masses per year if the process is assumed to have an efficiency of about 10 percent for producing energy from accreted mass. The rough estimate that 10 percent of the rest energy of the matter in an accretion disk would be eventually liberated as photons, in accordance with Einstein's formula E = mc², should be contrasted with a total efficiency of about 1 percent in nuclear reactions if a mass of hydrogen were to be converted entirely into iron. If the large-scale annihilation of matter and antimatter is excluded from consideration, the release of gravitational binding energy when matter settles onto compact objects is the most powerful mechanism for generating energy in the known universe. (Even supernovas use this mechanism, for most of the energy released in the explosion comes from the gravitational binding energy or mass deficit of the remnant neutron star.)

Interacting and merging galaxies provide the currently preferred routes to supply the matter swirling into the black hole. The direct ingestion of a gas-rich galaxy yields an obvious external source of matter, but the enhanced accretion of the parent galaxy's internal gas through tidal interactions (or bar formation) may suffice in most cases. At lower luminosities, other contributing factors may come from the tidal breakup of stars passing too close to the central black hole or from the mass loss from stars in the central regions of the galaxy. Gathering matter at a rate of two solar masses per year (90 percent of which ends up as the gravitating mass of the black hole) will build up a black hole of 10⁸ solar masses in several tens of millions of years. This estimate for the lifetime of an active galactic nucleus is in approximate accord with the statistics of such objects. This does not imply that supermassive black holes at the centres of galaxies necessarily accumulate from a seed of very small mass by steady accretion. There remain many viable routes for their formation, the study of such processes being in a state of infancy.  

Observational tests.

Except for the largest black holes or the nearest galaxies, the region interior to the turnup point is not resolvable by ground-based optical telescopes, because of the blurring effects produced by turbulence in the Earth's atmosphere. Excess central starlight and velocity dispersions have been seen in M87--a giant elliptical with a well-known optical jet emerging from its nucleus, which is located in the Virgo cluster, the nearest large cluster of galaxies. The excesses are consistent with a central black hole of several times 10⁹ solar masses. Atmospheric blurring, however, prevents astronomers from determining whether the upturns represent true cusps or merely shoulders that taper to constant values. Mere shoulders could be explained, without invoking a black hole, by the stars in the central regions of this galaxy having a nonstandard distribution of random velocities.

A better situation exists for the detection of supermassive black holes in the nuclei of spiral galaxies, since the interpretation of organized rotational motions is simpler than that for disorganized random motions. The Andromeda galaxy has an excess component of light within a few light-years of its centre. High-resolution spectroscopy of this region shows a large velocity width indicative of the presence of a black hole in the nucleus with a mass in excess of 10⁷ solar masses. Similar observations carried out for more distant spiral galaxies have yielded good candidates for supermassive black holes with masses ranging up to 10⁹ solar masses.

The closest galactic nucleus of all is of course located at the centre of the Milky Way Galaxy. Unfortunately, the nucleus of the system is not observable at the wavelengths of visible light, ultraviolet light, or soft X rays (those of lower energy than hard X rays), because of the heavy absorption by intervening dust. It can be probed by radio, infrared, hard X-ray, and gamma-ray techniques; such studies have revealed many intriguing features.

The most likely candidate for the nucleus of the Galaxy has long been regarded to be a compact radio-continuum source denoted Sagittarius A*. This synchrotron-radiation source is unique in the Galaxy: it is variable on a time scale of one day, implying that the radio emission arises from a region with dimensions smaller than the solar system; it shows evidence for synchrotron self-absorption, a condition consistent with a region being compactly filled with relativistic particles and fields; and measurements obtained with VLBI indicate that its motion with respect to the centre of the Galaxy is less than 40 km/sec, consistent with a heavy object brought to rest by "dynamic friction" in the deepest part of the Galaxy's potential well. Hard X-ray observations of the galactic central region, however, reveal only low-level emission from a diffuse component and several discrete sources with characteristics similar to coronal emission from luminous young stars. Broadband, near-infrared measurements at a wavelength centred near 2 micrometres (0.002 millimetre) show the presence of a dense star cluster. Surprisingly, the maximum concentration of light of the star cluster does not seem to centre on Sagittarius A*, nor does it show the r^-7/4 light cusp expected for the distribution of stars surrounding a massive pointlike object. Perhaps the cluster appears only by chance projection against the radio source. 

Spectroscopic investigations of the molecular and ionized gas yield a more promising interpretation. Molecular gas in a tilted ring within several light-years of the galactic centre exhibits rotational velocities consistent with motion under a central force field of an object having a mass of several million solar masses. Unfortunately, the molecular gas disappears before the centre can be approached very closely; fortunately, its disappearance is compensated by the appearance of ionized gas forming a "mini-spiral" within the central few light-years. One of the three arms of the mini-spiral streams within one light-year of Sagittarius A*. If this streamer is modeled as an infalling parabolic trajectory, a value of 4 10⁶ solar masses is obtained for a compact object at the nucleus of the Galaxy. If the Galaxy has a central black hole, this is probably the best estimate of its mass. 

Radio-continuum studies on a scale of hundreds of light-years from the Galaxy's centre show the nucleus to be embedded in an extraordinary set of filamentary arcs that pass perpendicularly through the galactic plane. Magnetic fields 1,000 times stronger than the general galactic field may play a role in defining the filaments, perhaps in a fashion analogous to the eruption of solar prominences. These magnetic fields may also have restrained the unusual massive molecular clouds Sagittarius A and Sagittarius B2 from forming OB stars with the same vigour as their counterparts farther out in the disk. Details such as these can be seen only because the nucleus of the Galaxy is so close (a "mere" 30,000 light-years away). This complexity should serve as a sobering reminder that most theoretical models of the active nuclei of external galaxies must vastly oversimplify the actual state of affairs. 

OTHER COMPONENTS

Cosmic rays and magnetic fields.

It is now known that cosmic rays come with both signs of electric charge and with a wide distribution of energies. About 83 percent of the positively charged component of cosmic rays consists of protons, the nuclei of hydrogen atoms, and about 16 percent of alpha particles, the nuclei of helium atoms. The nuclei of heavier atoms occur roughly in their cosmic abundances except that the light elements lithium, beryllium, and boron--which are quite rare elsewhere in the universe--are vastly overrepresented in the cosmic rays. The negatively charged component consists of mostly electrons at a level of 1 percent of the protons. Positrons also can be found, approximately 10 percent as frequently as electrons. A very small contribution from antiprotons is also known. Cosmic-ray positrons and antiprotons are believed to be by-products of collisions between the nuclei of cosmic rays with the ambient atomic nuclei that exist in interstellar gas clouds. Cosmic gamma rays, which have been detected emanating from the Milky Way and show a strong correlation with the distribution of interstellar gas, are another manifestation of such collisions. 

The cosmic-ray protons that freely enter the solar system, despite the outward sweep of the solar wind and the magnetic fields it carries, have energies that vary from a few times their rest energies to 10⁶ times and more. Thus, these particles must move at speeds approaching the speed of light. In this range the number of particles at energy E varies with E to the negative 2.7 power. A similar decreasing power law seems to hold for cosmic-ray electrons with energies from a few thousand to tens of thousands times their rest energies. Within uncertainties this energy distribution is consistent with the synchrotron-radiation interpretation of the nonthermal radio emission from the Galaxy. At higher energies, there are fewer cosmic-ray electrons than predicted by extrapolation of the power law found at lower energies, and this depletion can be understood on the basis of the large synchrotron-radiation losses suffered by the most energetic electrons.

Above 10⁷ times the rest energy of the proton, there also are fewer positively charged particles than predicted by the extrapolation of the power law E^-2.7; however, synchrotron losses cannot account for this deficiency. A more likely interpretation is that the cosmic-ray nuclei of lower energies are commonly produced and confined to the Galaxy, whereas those with very high energies may have an origin in very exotic or even extragalactic objects. This is consistent with the fact that protons with energies less than 10⁷ times their rest energies would be bent by the interstellar magnetic field to follow spiraling trajectories that would be confined to the thickness of the galactic disk. Nevertheless, these particles can eventually escape from the disk if the magnetic fields buckle out of the galactic plane (as they do because of certain instabilities).

An estimate of the total residence time of cosmic-ray nuclei within the disk of the Galaxy can be obtained by examining the anomalous abundances of lithium, beryllium, and boron. These elements are only somewhat less abundant in cosmic rays than carbon, nitrogen, and oxygen, and this has been conventionally interpreted to mean that the former group was mostly produced by spallation reactions (breakup of heavier nuclei) of the latter group as the cosmic-ray particles traversed interstellar space and interacted with the matter there. From the amount of spallation that has occurred, it can be estimated that the cosmic rays reside, on average, roughly 10⁷ years among the gas clouds in the galactic disk before escaping.

The origin of cosmic rays is an incompletely resolved problem. At one time astronomers believed that all cosmic rays, except those at the highest energies, originated with supernova explosions. The total energetics is right, and the presence in cosmic rays of nuclei as heavy as iron, etc., could receive a natural explanation under the supernova hypothesis. Unfortunately, doubt was cast on the hypothesis by later work that questioned, first, whether particles could really be accelerated to cosmic-ray energies in a single supernova shock and, second, whether these particles, even if accelerated, could propagate through the interstellar medium very far from the site of the original explosion. The second objection also applies to other possible point sources, such as pulsars.

A more promising possibility seems to be the proposal that cosmic rays are accelerated to their high energies by repeated reflections in magnetic shock waves in the interstellar medium (whose ultimate energy may be derived from the ensemble of all supernova explosions). The idea is that gas and the magnetic field threading it move at very different speeds on the two sides of the front of a shock wave. Cosmic-ray particles rattling through magnetic inhomogeneities may be shuttled back and forth between these two regions, gaining statistically an extra boost in energy every time they "bounce" off the moving set of magnetic field lines. The process is akin to the increasing energy that would be gained by a tennis ball in the absence of air drag if it were banged back and forth between a vigorously swinging player and a stationary wall. The great attractiveness of the strong shock-wave picture for accelerating cosmic rays is that it automatically gives, in the simplest models, a decreasing power-law distribution of particle energies. The exponent is 2 instead of the measured 2.7, and the discrepancy is believed to be related to an energy-dependent escape rate from the region of acceleration. The enhancement of the escape rate with increasing energy is not completely understood, but no fundamental obstacle appears likely in this direction to rule out the shock-acceleration model.

More serious failings of the shock-acceleration model are that it does not address the acceleration of cosmic-ray electrons, nor does it easily explain the origin of ultra-high-energy cosmic rays, nuclei with energies that lie between 10⁸ and 10¹¹ times the rest energy of the proton. There is some indication from measurements of ultra-high-energy gamma rays from some binary X-ray sources that these objects may copiously produce ultra-high-energy cosmic rays, but the exact acceleration mechanism remains obscure. At the highest observed cosmic-ray energies, the particles arrive preferentially from northern galactic latitudes, a fact interpreted by some to indicate a large contribution from the Virgo supercluster. In this picture even higher-energy cosmic rays from more distant parts of the universe (greater than about 10⁸ light-years) do not reach the Earth, because such particles would suffer serious losses en route as they interact with the photons of the cosmic microwave background. 

Microwave background radiation.

Interest in these calculations waned among most astronomers when it became apparent that the lion's share of the synthesis of elements heavier than helium must have occurred inside stars rather than in a hot big bang. In the early 1960s physicists at Princeton University, N.J., as well as in the Soviet Union, took up the problem again and began to build a microwave receiver that might detect, in the words of the Belgian cleric and cosmologist Georges Lemaître, "the vanished brilliance of the origin of the worlds."

The actual discovery of the relict radiation from the primeval fireball, however, occurred by accident. In experiments conducted in connection with the first Telstar communication satellite, two scientists, Arno Penzias and Robert Wilson, of the Bell Telephone Laboratories, Holmdel, N.J., measured excess radio noise that seemed to come from the sky in a completely isotropic fashion. When they consulted Bernard Burke of the Massachusetts Institute of Technology, Boston, about the problem, Burke realized that Penzias and Wilson had most likely found the cosmic background radiation that Robert H. Dicke, P.J.E. Peebles, and their colleagues at Princeton were planning to search for. Put in touch with one another, the two groups published simultaneously in 1965 papers detailing the prediction and discovery of a universal thermal radiation field with a temperature of about 3 K.

Precise measurements made by the Cosmic Background Explorer (COBE) satellite launched in late 1989 determined the spectrum to be exactly characteristic of a blackbody at 2.735 K. The velocity of the satellite about the Earth, the Earth about the Sun, the Sun about the Galaxy, and the Galaxy through the universe actually makes the temperature seem slightly hotter (by about one part in 1,000) in the direction of motion rather than away from it. The magnitude of this effect--the so-called dipole anisotropy--allows astronomers to determine that the Local Group of galaxies is moving at a speed of about 600 km/sec in a direction that is 45 from the direction of the Virgo cluster of galaxies. Such motion is not measured relative to the galaxies themselves (the Virgo galaxies have an average velocity of recession of about 1,000 km/sec with respect to the Milky Way system) but relative to a local frame of reference in which the cosmic microwave background radiation would appear as a perfect Planck spectrum with a single radiation temperature.

The origin of the "peculiar velocity" of 600 km/sec for the Local Group presents an interesting problem. A component of this velocity may be induced by the gravitational attraction of the excess mass above the cosmological mean represented by the Virgo cluster; however, it is now believed that the Virgo component is relatively small, at best 200-300 km/sec. A more important contribution may come from the mass of a "Great Attractor" at a distance of 10⁸ light-years connected to the Local Supercluster, but this interpretation is somewhat controversial since much of the supposed grouping lies behind the obscuration of the plane of the Milky Way. In any case, the generation of the large peculiar velocity of the Local Group of galaxies probably requires invoking an augmentation in dark matter of the gravitational attraction of the observable galaxies by a factor of roughly 10.

The COBE satellite carried instrumentation aboard that allowed it to measure small fluctuations in intensity of the background radiation, not just in the sense of a forward-backward asymmetry but also on angular directions in the sky that correspond to distance scales on the order of 10⁹ light-years across (still larger than the largest material structures seen in the universe, such as the enormous grouping of galaxies dubbed the "Great Wall"). The satellite transmitted an intensity pattern in angular projection at a wavelength of 0.57 centimetre after the subtraction of a uniform background at a temperature of 2.735 K. Bright regions at the upper right and dark regions at the lower left showed the dipole asymmetry. A bright strip across the middle represented excess thermal emission from the Milky Way. To obtain the fluctuations on smaller angular scales, it was necessary to subtract both the dipole and the galactic contributions. The latter requires a good model for the radio emission from the Galaxy at the relevant wavelengths, for which astronomers possess only incomplete knowledge. Fortunately, the corrections at high galactic latitudes are not very large, and an image was obtained showing the final product after the subtraction. Patches of light and dark represented temperature fluctuations that amount to about one part in 100,000--not much higher than the accuracy of the measurements. Nevertheless, the statistics of the distribution of angular fluctuations appeared different from random noise, and so the members of the COBE investigative team believe that they have found the first evidence for the departure from exact isotropy that theoretical cosmologists have long predicted must be there in order for galaxies and clusters of galaxies to condense from an otherwise structureless universe.

Apart from the small fluctuations discussed above (one part in 100,000), the observed cosmic microwave background radiation exhibits a high degree of isotropy, a zeroth order fact that presents both satisfaction and difficulty for a comprehensive theory. On the one hand, it provides a strong justification for the assumption of homogeneity and isotropy that is common to most cosmological models. On the other hand, such homogeneity and isotropy are difficult to explain because of the "light-horizon" problem. In the context of the cosmic microwave background, the problem can be expressed as follows. Consider the background radiation coming to an observer from any two opposite sides of the sky. Clearly, whatever are the ultimate sources (hot plasma) of this radiation, the photons, traveling at the speed of light since their emission by the plasma, have only had time to reach the Earth now. The matter on one side of the sky could not have had time to have "communicated" with the matter on the other side (they are beyond each other's light horizon), so how is it possible (with respect to an observer in the right rest frame) that they "know" to have the same temperature to a precision approaching one part in 100,000? What accounts for the high degree of angular isotropy of the cosmic microwave background? Or, for that matter, for the large-scale distribution of galaxies? As will be seen below in the section Cosmological models, a mechanism called "inflation" may offer an attractive way out of this dilemma. 

Intergalactic gas.

Very hot gas that emits X rays at tens to hundreds of millions of kelvins does indeed reside in the spaces between galaxies in rich clusters, and the amount of this gas seems comparable to that contained in the visible stars of the galaxies; however, because rich clusters are fairly rare in the universe, the total amount of such gas is small compared to the total mass contained in the stars of all galaxies. Moreover, an emission line of iron can frequently be detected in the X-ray spectrum, indicating that the intracluster gas has undergone nuclear processing inside stars and is not of primordial origin.

About 70 percent of the X-ray clusters show surface brightnesses that are smooth and single-peaked, indicative of distributions of hot gas that rest in quasi-hydrostatic equilibrium in the gravitational potentials of the clusters. Analysis of the data in the better-resolved systems allows astronomers to estimate the total amount of gravitating mass needed to offset the expansive pressure (proportional to the density times the temperature) of the X-ray-emitting gas. These estimates agree with the conclusions from optical measurements of the motions of the member galaxies that galaxy clusters contain about 10 times more dark matter than luminous matter (see below).

About half of the X-ray clusters with single-peaked distributions have bright galaxies at the centres of the emission. The high central densities of the gas imply radiative cooling times of only 10⁹ years or so. As the gas cools, the central galaxy draws the material inward at inferred rates that often exceed 100 solar masses per year. The ultimate fate of the accreted gas in the "cooling flow" remains unclear.

Another exciting discovery has been the detection of large clouds of atomic hydrogen gas in intergalactic space unassociated with any known galaxies. These clouds show themselves as unusual absorption lines in the Lyman-alpha transition of atomic hydrogen when they lie as foreground objects to distant quasars. In a few cases they can be mapped by radio techniques at the spin-flip transition of atomic hydrogen (redshifted from the rest wavelength of 21 centimetres). From the latter studies, some astronomers have inferred that the clouds exist in highly flattened forms ("pancakes") and may contain up to 10¹⁴ solar masses of gas. In one interpretation these structures are the precursors to large clusters of galaxies (see below). 

Low-energy neutrinos.

Such low-energy neutrinos, nonetheless, attracted considerable astronomical interest during the late 1970s because experiments conducted in the Soviet Union and the United States suggested, contrary to the prevailing belief in particle physics, that neutrinos may possess a nonzero rest mass. Even if the rest mass were very small--say, 10,000 times smaller than the rest mass of the electron, the lightest known particle of matter--the result could be of great potential importance because neutrinos, being so relatively abundant cosmologically, could then be the dominant source of mass in the universe. Unfortunately, later experiments cast doubts on the conclusions of the earlier findings, and theoretical investigations of "massive neutrinos" as the dark matter in the universe turned up as many new difficulties to be explained as possible solutions to old problems. On the other hand, if the solution to the solar-neutrino problem turns out to depend on the existence of neutrino oscillations, massive-neutrino cosmologies may well make a (partial) comeback. 

Gravitational waves.

There are, however, some indirect pieces of evidence that accelerated astronomical masses do emit gravitational radiation. The most convincing concerns radio-timing observations of a pulsar located in a binary star system with an orbital period of 7.75 hours. This object, discovered in 1974, has a pulse period of about 59 milliseconds that varies by about one part in 1,000 every 7.75 hours. Interpreted as Doppler shifts, these variations imply orbital velocities on the order of 1/1000 the speed of light. The non-sinusoidal shape of the velocity curve with time allows a deduction that the orbit is quite noncircular (indeed, an ellipse of eccentricity 0.62 whose long axis precesses in space by 4.2 per year). It is now believed that the system is composed of two neutron stars, each having a mass of about 1.4 solar masses, with a semimajor axis separation of only 2.8 solar radii. According to Einstein's theory of general relativity, such a system ought to be losing orbital energy through the radiation of gravitational waves at a rate that would cause them to spiral together on a time scale of about 3 10⁸ years. The observed decrease in the orbital period in the years since the discovery of the binary pulsar does indeed indicate that the two stars are spiraling toward one another at exactly the predicted rate.

The implosion of the core of a massive star to form a neutron star prior to a supernova explosion, if it takes place in a nonspherically symmetric way, ought to provide a powerful burst of gravitational radiation. Simple estimates yield the release of a fraction of the mass-energy deficit, roughly 10⁵³ ergs, with the radiation primarily coming out at wave periods between the vibrational period of the neutron star, approximately 0.3 millisecond, and the gravitational-radiation damping time, about 300 milliseconds.

A cosmic background of gravitational waves is a possibility that has sometimes been discussed. Such a background might be generated if the early universe expanded in a chaotic fashion rather than in the smooth homogeneous fashion that it is currently observed to do. The energy density of the gravitational waves produced, however, is unlikely to exceed the energy density of electromagnetic radiation, and each graviton (the gravitational analogue of the photon) would be susceptible to the same cosmological redshift by the expansion of the universe. A roughly thermal distribution of gravitons at a present temperature of about 1 K would be undetectable by foreseeable technological developments in gravitational-wave astronomy. 

Dark matter.

Another possibility is that the dark matter is (or was) composed of ordinary matter at a microscopic level but is essentially nonluminous at a meaningful astronomical level. Examples would be brown dwarfs (starlike objects too low in mass to fuse hydrogen in their interiors), dead white dwarfs, neutron stars, and black holes. If the objects are only extremely faint (e.g., brown dwarfs), they can eventually be found by very sensitive searches, perhaps at near-infrared wavelengths. On the other hand, if they emit no light at all, then other strategies will be needed to find them--for example, to search halo stars for evidence of "microlensing" (i.e., the temporary amplification of the brightness of background sources through the gravitational bending of their light rays).