Proc. Nat. Acad. Sci. USA Vol. 72, No. 12, pp..4685-4690, December 1975 H. P. Robertson Memorial Lecture Progress and prospects in high energy astrophysics* MARTIN J. REES Institute of , Madingley Road, Cambridge, England The traditional kind of astrophysicist is, in a sense, an "ap- ; a compact neutron spinning 30 times a second, plied" physicist, who computes models for and emitting a beam of radio and optical emission which swings based on relatively well-understood properties of atoms and through our line of sight once per revolution. Although neu- nuclei, Newtonian , and other branches of classical tron stars remained hypothetical objects until the discovery physics. But recently radio and x-ray observations have re- of in 1968, their properties had been discussed by vealed some fascinating cosmic objects and phenomena Landau, Zwicky, Oppenheimer, Volkoff, and others in the where the inferred energies, densities, and gravitation field 1930s. Neutron stars are estimated to have masses of the strengths are so extreme that we cannot be confident that we order of that of the , but radii of only 10 or 20 km. From know the relevant physics. The physical assumptions them- these numbers one can estimate their moment of inertia. selves, and not merely the astrophysical models, are then The Crab pulsar is observed to be slowing down, implying vulnerable to observational test; and the astrophysicist can that rotational energy is being lost; and, as was first empha- feel that he has a symbiotic rather than a parasitic relation- sized by Gold, the rate of loss of kinetic energy was just suf- ship with his physicist colleagues. My aim today will be to il- ficient, if channeled efficiently into relativistic particles, to lustrate this theme by describing, in an inevitably sketchy provide the inferred power input into the . way, some current ideas on neutron stars, black holes, and The discovery of pulsars has stimulated a great deal of de- the early universe. tailed study of the physics of nuclear matter and neutron stars. A slice through a would in some respects NEUTRON STARS AND PULSARS look rather like a slice through the Earth, in that it would I shall lead into this subject in what may seem to the astrono- display a crust, a liquid interior, and maybe a solid core (ii). mers present to be a very hackneyed way, by discussing the The outer crust would consist mainly of 56Fe, but at higher Crab Nebula-the expanding debris of a explo- densities the main constituents would be progressively more sion witnessed and recorded by Chinese and Japanese as- exotic neutron-rich nuclei such as 62Ni, 82Ge, 124Nb, and tronomers in the year 1054 A.D. Supernova explosions sig- "18Kr. Though unstable in the laboratory, these nuclei are nify the violent endpoint of , when a star too prevented from decaying by the dense Fermi sea of elec- massive to become a exhausts its available nu- trons. The nuclei would be in a crystalline lattice. As one clear energy. The core catastrophically implodes, and (by digs still deeper below the crust one reaches densities ap- processes whose details are poorly understood) the outer proaching 1014 g cm-3. Under these conditions the matter is layers are blown off. Filamentary ejecta such as are seen in mainly in the form of free neutrons. Moreover the neutrons the Crab Nebula play a crucial role in the "ecology" of the display superfluid properties, analogous to those displayed , because the heavy elements in all stars like the Sun by terrestrial at temperatures of a few millidegrees. are believed to have been synthesized in stars that exploded The small fraction of the material in the form of protons early in galactic history, the having condensed (and perhaps the electron gas as well) would constitute a su- from interstellar gas already contaminated by debris from perconductor. In the core of a neutron star, where the densi- such explosions. Supernovae are an important heat input ty may attain values of 1015 g cm-3, the character of the into the interstellar medium and are also believed to be per- material is uncertain, because normally unstable hyperons haps the predominant regions where galactic cosmic rays are may become quite stable under these conditions. Also, it is accelerated. still an important and controversial issue whether, at these The inner part of the Crab Nebula, and the space between extreme densities and pressures, a strongly interacting sys- the filaments, emits polarized continuum radiation, not only tem of baryons can turn into a nuclear solid. in the optical band, but also in the radio and x-ray parts of At the first sight one might fear that little could be the spectrum. (In this respect, the Crab Nebula is morpho- learned about the interior structure of a neutron star by as- logically different from the remnants of several other "his- tronomical observations. However, the timing precision of torical" supernovae-e.g., Tycho's supernova of 1572 and the Crab pulsar is so great that it is possible to detect very Kepler's supernova of 1604-which have a shell structure.) small irregularities in the rate at which it is slowing down. The continuum is interpreted as synchrotron radiation, On at least one occasion, in fact, it appears that the period implying the presence of relativistic electrons with energies underwent a sudden slight decrease. The most natural inter- up to 105 GeV and magnetic fields of a few times 10-4 G (a pretation of this is as follows. As a neutron star spins down it few times 10-8 tesla). The calculated synchrotron lifetimes adjusts its shape from an oblate spheroid to an increasingly are such that these electrons cannot all be survivors of the spherical configuration. (A rotation rate of >1000 revolu- explosion 900 years ago, but must be continually injected tions per second is required for centrifugal breakup, so even into the Nebula. How this could have occurred was a mys- for the Crab pulsar-the one with the shortest known pe- tery until the discovery that the faint central star, which ap- riod-the equatorial flattening is less than 1 part in 103.) If pears quite ordinary on a photographic plate, was actually a t This book contains several detailed up-to-date reviews relevant to neutron star theory and is recommended as a general source of * H. P. Robertson Memorial Lecture presented at the National further information on other topics briefly mentioned in this lec- Academy of Sciences, Washington, D.C., April 1975. ture. 4685 Downloaded by guest on September 27, 2021 4686 Robertson Memorial Lecture: Rees Proc. Nat. Acad. Sci. USA 72 (1975)

the star were a fluid, this adjustment would take place con- fields, providing conditions that are propitious for accelerat- tinuously. However, since the crust (and possibly also the ing particles to high individual energies. core) of a neutron star can maintain a certain amount of All other known pulsars are spinning more slowly than the stress, this readjustment takes place in jerks, causing sudden Crab pulsar; they are presumably older, the remnants of the decreases in the moment of inertia. Such is the precision of supernovae that gave birth to them having long ago dis- the observations that a decrease in the equatorial radius of a persed and merged into the diffuse interstellar medium. neutron star by only about 10 gm is readily detectable in the From the estimated number of pulsars in the Galaxy, one timing data. There is some evidence for very small but per- can infer that a substantial fraction of supernovae must form sistent random jitter in the period of the Crab pulsar, and spinning neutron stars. Some supernovae, however, probably this is attributed by some theorists to "microquakes" in- do not; the fact that the remnants of the other "historical" duced by misalignment of the rotation axis and the symme- supernovae look so different from the Crab suggests that try axis-analogously to the Earth's Chandler wobble. Thus, these do not contain a continuing central source of power studies of these inferred starquakes can in principle provide injecting relativistic electrons into them. some of the same information about the interior of neutron It is plainly of great interest to determine the masses of stars as seismology provides about the interior of the Earth; neutron stars. So far, this has not been possible for any pul- it is perhaps in this way that we can best investigate the be- sar. However, two well-studied x-ray sources, Her X-1 and havior of matter at supernuclear densities. Cen X-3, are generally interpreted as neutron stars orbiting Neutron stars also appear to resemble the Earth in that around a (relatively ordinary) companion star, and in these they possess magnetic fields and a magnetosphere: the sur- cases one can in principle obtain mass estimates from the face magnetic field strength is about 1012 G (about 108 Newtonian dynamics of the orbit. In these objects, the emis- tesla). Processes occurring in the magnetosphere are be- sion mechanism for the x-rays is very different from that in lieved to be responsible for the pulsed radiation that we see the Crab pulsar. Instead of being associated with an electro- from pulsars, and also for accelerating the relativistic parti- magnetically-driven relativistic outflow powered by the ro- cles that keep the Crab Nebula shining in the optical and tational kinetic energy, the x-rays from Her X-1 and Cen x-ray continuum. (The magnetic field would be strong X-3 represent thermal radiation by the neutron star surface, enough to enforce corotation of the magnetosphere almost which is heated by the impact of infalling gas accreted from out to the "light cylinder" where the corotation velocity ap- the companion star. proaches c, and in this situation the acceleration of particles Before finally leaving the Crab Nebula, I might venture to ultra-relativistic energies can be efficient.) Neutron stars one historical speculation. The Crab pulsar can be detected exemplify supremely well how astronomical observations visually through any large telescope, but its pulse repetition allow us to study material under extreme conditions that rate is so high that the eye responds to it as a steady source. could not possibly be simulated terrestrially. I have already But had it been spinning say 10 times a second, rather than alluded to the extremes of density and pressure which they 30 times, the remarkable properties of the central star of the involve. The radiation densities in pulsars are also extreme, Crab Nebula could have been discovered 50 years ago. How and may exceed by a substantial factor those in terrestrial would the course of 20th century physics have been changed laser beams. The magnetic fields are a million times stronger if superdense matter had been detected in the 1920s? One than the strongest laboratory fields. Indeed, a theory of pul- cannot guess, except that the importance of astronomy for sar emission proposed by Sturrock (2), and recently elabo- fundamental physics would surely have been recognized far rated by Ruderman and Sutherland (3), suggests that pro- sooner. duction of electron-positron pairs by photons moving transversely to these fields-a process never observable in BLACK HOLES the laboratory-is directly responsible for the pulsed emis- Now there is another class of objects in which gravity could sion. These magnetic fields are so strong that, instead of be even more dominant than in neutron stars. These objects being merely a perturbation, they completely dominate the are "black holes"-bodies that have collapsed to such small structure of atoms exposed to them: in a 1012 G (108 tesla) dimensions that no light nor any other signal can escape field, even the inner shell electrons of iron are so severely af- from them. The possible occurrence of black holes is in fact fected by magnetic effects that the atoms are essentially a consequence of almost all theories of gravitation, not mere- "one-dimensional" in character. ly of Einstein's general relativity. Indeed, their existence Neutron stars are also extreme in the strength of their gra- was, in essence, conjectured by Laplace in 1798 on the basis vitational fields. The acceleration due to gravity at their sur- of Newtonian gravity and the ballistic theory of light. It is of faces is 1013 times that on Earth. The gravitational binding course clear that Newtonian gravity is not appropriate for energy is typically 10% of the rest mass energy, implying discussion situations where the gravitational fields are very that the escape velocity from their surface is about one-third strong. But it is still not completely evident which theory of the speed of light. Indeed, it is from gravitation that the should be used to extend it (even though Einstein's theory is magnetic and rotational energy of pulsars is derived. If an apparently being vindicated-at least in the "post-Newto- ordinary implodes to neutron star densities, con- nian" approximation-by increasingly precise experimental serving angular momentum, then its rotational energy in- tests). creases. In an analogous way an ordinary stellar magnetic The first theoretical description of a within the field can be amplified to 1012 G (108 tesla), if flux is con- framework of general relativity was given in 1917 by Karl served during the compression. Thus, gravitation cannot Schwarzschild, who calculated the gravitational field around only energize the supernova explosion itself, but can also a spherically symmetrical mass. The so-called Schwarzschild leave a "dead" star with more energy than nuclear reactions metric has the property that there is a horizon at a radius 2 could generate over its entire previous lifetime. Moreover, GM/c2, or approximately 3 M km, where M is measured in much of the gravitational energy goes into the organized solar masses (which is in fact exactly the same as Laplace low-entropy form of rotation and large-scale magnetic calculated). This corresponds to the minimum radius from Downloaded by guest on September 27, 2021 Robertson Memorial Lecture: Rees Proc. Nat. Acad. Sci. USA 72 (1975) 4687

which light can escape to an external observer; or, almost But there are several other reasons for studying black equivalently, the radius at which the gravitational redshift holes. They interest the astrophysicist because they are in a factor is infinite and the escape velocity equal to c. Whereas sense the "ghosts" of dead stars and tell us something about the gravitational deflection of light is a very small effect in the last stages of stellar evolution; they are objects that have the solar system and in most astronomical contexts, light is collapsed, cutting themselves off from the rest of the Uni- severely bent in the strongly curved space close to the verse, but leaving a gravitational imprint frozen in the space Schwarzschild radius Rs. An experimenter situated just out- they have left. To the physicist gravitational collapse is im- side Rs would need to aim a beam of light almost directly portant, because the singularity must be a region where the outwards in order that it should not be dragged back and laws of classical gravitation break down and one needs to in- eventually swallowed by the black hole: if he ventured with- voke some fully quantum theory of gravity to understand in Rs, he would be unable to send any light signals to the ex- what is going on. Many people have claimed that the para- ternal world. Although the region enclosed within the doxes associated with the singularity within a black hole are "Schwarzschild horizon" is shrouded from an external ob- as fundamental, and as far reaching in their implications, as server's view, a freely falling experimenter could pass inside the puzzles connected with black body radiation and the sta- it without experiencing anything especially unusual as he bility of bound electron orbits in atoms, which confronted went through the critical radius. He would then, however, physicists at the beginning of the 20th century and triggered have entered a region from which, however hard he acceler- the development of the quantum theory. ated, he could never escape. He would eventually reach a Black holes also have an important bearing on our general central singularity where tidal forces (the difference be- concept of space and time, because near them space behaves tween the gravitational acceleration of his hands and his in peculiar and highly "non-intuitive" ways. For instance, feet) would become infinite, and he would be crushed out of time would "stand still" for an observer who managed to existence. The infalling experimenter would reach this so- hover just outside the horizon, and he could see the whole called singularity in a definite finite time of order RS/c as future of the external universe in what, to him, was quite a measured by his own clock. But an external observer would short period; and even stranger things might happen if one never see him fall within the Schwarzschild radius: as the ventured inside the horizon. horizon was approached, the falling experimenter's clock But would one really expect black holes to exist? Most as- would appear to the distant observer to run slower and slow- trophysicists would say yes for the following reasons. Over er, and any signal that he sent would become more and 40 years ago Chandrasekhar (5) calculated an upper limit to more redshifted. The redshift in fact exponentiates on a ti- the mass of a white dwarf star. His conclusion was specially mescale comparable with the light travel time across the startling because the calculated upper limit was little more Schwarzschild radius. For a black hole, this time than one , which immediately raised the question is very short, implying that the experimenter would disap- as to what eventually happened to stars of much greater pear and his signals fade in a fraction of a second. Thus one mass when they had exhausted their nuclear energy. It is of cannot from a safe distance observe anything about the ex- course possible that a star in the course of its evolution can treme conditions near the central singularity. One's Faustian lose most of its mass, so that eventually it settles down as a urge must be sufficiently strong to venture inside the hori- white dwarf of mass below the . Anoth- zon despite the inevitable destruction this implies! er somewhat less mundane possibility is that the outer layers In 1963 Kerr (4) discovered an exact solution of Einstein's may be ejected explosively in a supernova event at the end equations that was more general that the Schwarzschild met- of the star's lifetime, leaving a neutron star, as seems to have ric (but still possessed a "horizon" with radius about GM/c2) happened in the Crab Nebula. But theory now sets a limit to and which he believed represented the field around a col- the mass of a neutron star. This limit is a bit uncertain be- lapsed rotating object. One of the key results of recent theo- cause it depends on the equation of state for nuclear matter retical progress in general relativity has been the proof that, at densities about 1015 g cm-3, but is almost certainly below in contrast to the situation in Newtonian theory, a singular- three solar masses (and it is comforting that the best esti- ity still occurs even when the collapse has no special symme- mates of the mass of Her X-1 yields about 1.3 M). try. Moreover, it seems that once a black hole forms it emits It would seem unlikely that stars always display such a burst of gravitational waves and quickly settles down to a "prescience" that they shed enough gas to bring them safely standardized stationary state where its external gravitational. below this limit, and any stellar remnant more massive than field is characterized essentially by two free parameters, about two or three solar masses would seem fated to undergo mass and angular momentum. (In principle, electric charge complete gravitational collapse, because internal pressures is a third conserved quantity, but it is unlikely that any as- could not maintain it in equilibrium when its nuclear energy trophysically realistic collapse could generate a black hole sources were exhausted. Moreover, stars more massive than with significant charge.) This result is colloquially described two or three solar masses complete their evolution in a time as the theorem that "black holes have no hair." Thus the so- very short compared to the age of our Galaxy. Some types of called Kerr solution, once regarded as a special and perhaps supernovae may form black holes rather than neutron stars; atypical case, has now acquired paramount importance be- alternatively, some massive stars may undergo complete gra- cause it apparently describes the metric of space-time vitational collapse without producing any conspicuous su- around any black hole. Most other viable theories of gravity, pernova-type explosion. and in particular the widely discussed scalar-field theory of These considerations suggest that the Galaxy may indeed Brans and Dicke, also predict that black holes can form, contain large numbers of gravitationally collapsed bodies. though with slightly different properties; and a strong moti- How does one then search for them? One might, of course, vation for searching for black holes is that they represent detect the burst of gravitational waves signaling their forma- objects where gravity has overwhelmed all other forces, al- tion (which may or may not be associated with some spec- lowing one to test theories of gravitation under the most ex- tacular outburst of electromagnetic energy like a superno- treme conditions. va). But to do this one would &ave to be v-e-ry lucky: a factor Downloaded by guest on September 27, 2021 4688 Robertson Memorial Lecture: Rees Proc. Nat. Acad. Sci. USA 72 (1975) of 108 improvement over the sensitivity of Weber's detectors ometry, planetary radar, and the like-in the solar system, would be required before one could detect stellar deaths at even though there the non-Newtonian effects are very small such great distances-e.g., as far away as the Virgo Cluster indeed (and perhaps, also, by studying the orbit of the newly of galaxies-that the expected event rate exceeded a few per discovered ). But in order to study situations year. where gravity is overwhelmingly strong, one must look fur- Once formed, black holes are essentially "passive" and the ther afield, and the highly suggestive evidence already ob- best hope of locating them then lies in discerning their grav- tained will certainly stimulate, and should surely justify, an itational effects on the dynamics of neighboring stars or gas. intensive quest for black holes using all relevant techniques It was suggested by Zeldovich and others back in 1964 that of space and ground based astronomy. one might detect x-rays from black holes in close binary sys- tem, which are accreting gas from their companion. Cap- GALACTIC NUCLEI AND COSMOLOGY tured gas would form a disc, swirling downwards into the Pulsars and compact x-ray sources are among the most exot- potential well surrounding the black hole. Viscous dissipa- ic objects available for study within our own Galaxy, but tion would heat it to temperatures of 108 K so that its when we extend our horizons into the extragalactic realm we bremsstrahlung emission would be in the x-ray band. These observe many puzzling phenomena which appear to be asso- x-rays would display rapid irregular flickering. ciated with the nuclei of a certain class of galaxies. These The small NASA satellite called "Uhuru," launched in nuclei are sources of intense non-stellar radiation, signifying 1970, has led to several remarkable advances in our knowl- the presence of a highly concentrated energetic source gen- edge of the x-ray sky. But perhaps the most outstanding of erating (in many instances) more power than the whole of these has been the identification by Giacconi and his col- the rest of the Galaxy. An example of this phenomenon is leagues (6,*) of at least one source that fits this description. NGC 1275-a Seyfert galaxy. The optical appearance of this The x-ray source X-1 does indeed display irregular galaxy is strongly reminiscent of the Crab, though of course flickering and is in a very close-almost grazing-orbit the and scale is much larger. It seems that here around a companion star. Moreover, from the ordinary again we observe the effects of synchrotron radiation, Newtonian dynamics of the binary system the best estimate implying particle acceleration; and also we observe fila- of the x-ray object's mass is at least six solar masses, implying ments ejected at high speed. In fact, one reason why the that it cannot be a neutron star or white dwarf. The natural Crab Nebula is so important to the astrophysicist is that it conclusion seems to be that Cygnus X-1 involves a black hole permits him to study, at relatively close range within our of at least six solar masses. It is impossible to construct a own Galaxy, the same physical processes that may be operat- completely watertight case, since ad hoc and contrived mod- ing on a much grander scale in external galactic nuclei. els that do not involve a black hole can always be devised. Radio galaxies such as display this type of activ- One's assessment of the odds obviously depends on whether ity on an even more spectacular scale than NGC 1275. In one regards black holes as inherently absurd, or as plausible Cygnus A one infers the presence of a central source of rela- endpoints for stellar evolution. The case would be firmer if tivistic plasma, which is somehow channeled into two oppo- one had several examples of similar sources. It is possible sitely-directed beams. The radio emission comes predomi- that several of the sources already discovered by the Uhuru nantly from symmetrical "blobs" on either side of the gal- satellite will turn out to be systems similar to Cygnus X-1. axy, which probably correspond to places where the beams Moreover by 1980 NASA's High Energy Astronomical Ob- are currently encountering the intergalactic medium, and servatory x-ray telescope (HEAO B*), more than 103 times their energy is being converted into random motions of rela- more sensitive than those so far available, should be in orbit. tivistic particles which then radiate via the synchrotron pro- It will then be feasible to study the properties of the known cess (7). The disturbed-looking nearby galaxy Centaurus A sources in vastly greater detail and perhaps discover many was perhaps, 108-109 years ago, a powerful radio galaxy like more of the same kind, maybe even some in external galax- Cygnus A; but it is now a relatively weak and very diffuse ies. source, with only a low level of residual non-thermal activity Much better observations, and a great deal more theoreti- in its nucleus. The energies involved in these radio galaxies, cal model-building, will be needed before one can diagnose and also in the quasars, range up to 1060 ergs (1053 J)- whether the properties of a given x-ray source agree better equivalent to the rest mass of half a million stars like the with those expected on the basis of general relativity than Sun. with the predictions of an alternative theory of gravitation. An increasingly plausible hypothesis is that radio galaxies, Nevertheless the probable discovery of black holes does open Seyfert galaxies, and probably the quasars as well, are varied the way to testing some of the most crucial and remarkable manifestations of the same general kind of violent activity in predictions of Einstein's theory. galactic nuclei. Among the suggested models proposed to ex- The renaissance in gravitational research in the last dec- plain this general phenomenon are that one is seeing a rapid ade has been due partly to the utilization by the theorists of succession of supernova-type explosions, resulting in the for- new and more powerful topological techniques (which have mation of a million Crab Nebulae almost in unison. Alterna- led to the proof of the "no hair" theorems); but it has also tively it has been speculated that a single magnetized super- been stimulated by the realization that astrophysical phe- massive spinning object may be involved, which accelerates nomena may actually exist where relativistic effects are particles rather like a giant pulsar. A third possibility is that large. Theories of gravity can essentially be tested only by a massive black hole lurks in the center of these galaxies and astronomical observations and by experiments in a cosmic accretes gas from its surroundings rather like a scaled-up environment, rather than by laboratory work. In the short version of the x-ray source Cygnus X-1. run the most useful tests of rival gravitation theories may All these ideas invoke gravitation or gravitational contrac- well come from high precision experiments-radio interfer- tion as the primary energy source. But it is still an entirely open question which of these three alternatives (if any) is f See also the extensive discussions of Cyg X-1 in ref. 1. most likely to be correct. Our best observational line of at- Downloaded by guest on September 27, 2021 Robertson Memorial Lecture: Rees Proc. Nat. Acad. Sci. USA 72 (1975) 4689

tack on this problem will involve detailed optical studies of more indirect technique of estimating the mean density of quasars and the variable central regions of active galaxies; all forms of gravitating matter-galaxies, intergalactic gas, and investigations of the structure of compact radio sources collapsed objects, and other even more elusive possibilities. by the technique of very long baseline interferometry. An important impetus to relativistic cosmology stemmed Even though we understand very little about them, qua- from Penzias and Wilson's (9) discovery of the isotropic mi- sars and radio galaxies are astrophysically important for var- crowave background radiation, which was interpreted-fol- ious reasons. According to almost all the alternative theories lowing earlier theoretical ideas of Alpher, Gamow, and Her- the active lifetime of a quasar or radio source is very short man, and of Dicke-as a relic of the hot dense early phase of compared to the Hubble time, implying that there have the expansion. This discovery is regarded by most people as been very many generations of such objects; so the dead ones of comparable importance to Hubble's realization of the must greatly outnumber the living. This therefore means universal expansion. The so-called "hot big bang" theory, in that if we integrate over the whole lifetime of the Universe, its simplest form, predicts that this radiation should have a they have had an important impact on the energy budget of black body spectrum; and the evidence that this may indeed intergalactic space, and may have been responsible for heat- be the case, the present temperature being about 2.7 K, ing intergalactic gas and for producing any intergalactic flux has somewhat strengthened in recent months with the an- of cosmic rays. nouncements from London (10) and Berkeley (11,§) of two Quasars and radio galaxies-even if we do not understand direct measurements made from balloons in the crucial mil- their precise nature-are important also for observational limeter waveband where the spectrum has its peak intensity. cosmology (8), because they are hyperluminous beacons al- Whereas exploding galaxies and quasars may permit us to lowing us to probe deeper into space (in other words, further sample up to 90% of cosmic history since the initial big back into the past) than is possible by means of normal gal- bang, the microwave background radiation has been propa- axies. The most dramatic inference from these analyses is gating uninterruptedly through space for 99.99% of the time that the universe is by no means in a steady state, and that since the big bang. This provides direct evidence that the the density of powerful sources per comoving volume (i.e., Universe on a large scale is homogeneous and isotropic to a over and above the density enhancement factor resulting remarkable degree. It was of course H. P. Robertson, whom from the universal expansion) would have been about a we are commemorating by this lecture, who emphasized thousand times higher at early epochs than at the present that this uniformity restricts the admissible description on time. The most distant observed known sources-quasars the Universe to a very limited range of mathematically cal- with redshifts >-3.5-are so far away that they emitted the culable models. Were this not so, progress in -cosmology radiation that we now detect when the Universe was 20% of would be almost inconceivable. If one assumes that the Uni- its present "age" of ,1010 years. By observing such objects verse was isotropic and almost homogeneous right back to we are thus in effect probing 80% of cosmic history, and very early times, and assumes also that Einstein's relativity is maybe it should not surprise us when we infer such dramatic applicable, it is possible to calculate the chemical composi- differences between conditions at these ancient epochs and tion of the material emerging from the big bang; and it is at the present day. A hypothetical astronomer observing widely regarded as one of the triumphs of Gamow's theory only two billion years after the initial singularity would per- that it enables us to predict that this material will be 25% he- ceive a vastly more active and dramatic environment: lium and 75% . The relevant nuclear reactions all whereas our nearest bright quasar, 3C 273, is about 800 me- occur during the first -100 sec of the expansion. This result gaparsecs (Mpc) distant, he would be likely to find a similar is in gratifying apparent accordance with observations and object only 15 Mpc away-i.e., 50 times closer-and ap- solves a long-standing problem in the theory of nucleogen- pearing as bright as a fourth star. He would ob- esis; because, whereas the heavier elements can perhaps be serve correspondingly greater activity in the radio sky, the adequately explained by nucleosynthetic processes associ- integrated flux at meter wavelengths being perhaps a hun- ated with stellar evolution, it always proved a problem to ex- dred times higher than it is today. This evidence is obviously plain the high abundance and relative uniformity of helium. crucially important to our understanding of the astrophysi- There is of course no reason why we should stop our back- cal evolution of the contents of the universe. ward extrapolation at the time when the age of the Universe It is however disappointing that the discovery of these was 100 sec. The density would have attained nuclear densi- objects with very large redshifts has not lead to any progress ties at a time l0-5 sec after the big bang, and before this in "classical" or geometrical cosmology. But this problem time we encounter the same uncertainties about the behav- will remain until we understand enough about the physics of ior of matter as arise in the cores of neutron stars. But if, as these objects to be able to insert the appropriate evolutionary perhaps we should, envisage time on a logarithmic scale, to correction. All that can be said at the moment (and this was ignore the first 10-5 sec would seem a severe omission in- important in the early 1960s when the steady state theory, in deed! The early stages of the big bang are in many senses which no global evolution is permitted, was being actively analogous to the time-reversal of a gravitational collapse, discussed) is that evolutionary effects are certainly very and theory predicts that the densities and space curvatures large. We still do not know whether our universe is likely to should become arbitrarily high as we approach close to the continue expanding forever, or whether the mutual gravita- initial instant t = 0. When the space curvature becomes suf- tional attraction of all its constituents will cause an eventual ficiently great (t < 10-23 sec) it is believed that particle recollapse, during which everything will suffer the fate of pairs can be created out of the gravitational fields in an anal- the material in a black hole. Indeed, some astronomers sus- ogous manner to the creation of electron-positron pairs in pect that our best technique for tackling this issue is not by strong electromagnetic fields. When the curvatures reach extending Hubble's work to bigger redshifts and bigger even more extreme value (t < lo4 sec) it is accepted that look-back times (because, even though the effects of the de- celeration should then be larger, the various evolutionary corrections are also larger and more uncertain), but by the §See ref. 8 for full discussions of the microwave background. Downloaded by guest on September 27, 2021 4690 Robertson Memorial Lecture: Rees Proc. Nat. Acad. Sci. USA 72 (1975) Einstein's theory cannot be an adequate description and that In ground based and space astronomy alike the major a full-blown theory of quantum gravity is required. When a payoff and vindication of past projects has often lain in the stellar mass object collapses to become a black hole the re- entirely unforeseen range and richness of new phenomena gions of such extreme space curvature are deep inside the that they revealed. This discourages me from venturing any horizon, shrouded from view. But this is not so for the big specific predictions of future developments. In any case bang, where there is in principle a causal chain extending most people here have probably read the report published 3 right back to the singularity. Many relativists believe that years ago by Professor Greenstein and his colleagues (14) the explanation of the global isotropy and homogeneity of under the auspices of this Academy that contains a compre- the Universe must be sought in dissipative processes associ- hensive review of the whole present state of astronomy and ated with quantum gravitational effects. an inspiring manifesto for the future. Many branches of as- Hawking has recently conjectured (12, 13), incidentally, tronomy, particularly those dependent on space techniques, that if the very early Universe were irregular, the tremen- are still at the exciting pioneering stage where orders of dous pressures could have created small black holes of kinds magnitude improvement are feasible within the next few that could not be produced astrophysically at current ep- years. When one recalls the many new things that radio as- ochs. A black hole the mass of the Earth would have a tronomy has taught us in the last twenty years, and how dif- Schwarzschild radius of only 1 mm, and a mass of 1015 g ferent the radio sky is from that viewed through an optical would produce a black hole whose dimensions were compa- telescope, one can confidently expect further revolutionary rable with those of an elementary particle. In these "mini- surprises which may demonstrate how biased and incom- holes" the space curvature, even outside the horizon, is large plete our present picture of the Universe really is. The whole enough that particle creation from the gravitational field subject is one in which the ratio of problems to practitioners would be occurring. Such objects would actually emit ther- remains remarkably high by the standards of the physical mal radiation with a temperature depending inversely on sciences; and where, as the frontiers have advanced, their their mass. If these "evaporating" mini-holes actually existed periphery has dramatically enlarged. In a sense this aggrav- as "fossils" of the initial instants after the big bang, they ates the task of selecting those projects that can be supported might be detectable in the gamma ray band, and would be from limited resources. But those who have to make such "missing links" which might facilitate the long range quest choices in this country can perhaps feel gratified and en- for unification of particle physics and gravitation theory. couraged at having already fostered an astrophysics pro- gram whose achievements will surely seem in retrospect one CONCLUSIONS of the highlights of 20th century scientific endeavor. To summarize, my remarks have intentionally focused on 1. "Astrophysics and gravitation", Proc. 16th Solvay Conference some of the most extreme-even pathological-phenomena Brussels 1974. on the cosmic scene. But even though neutron stars, black 2. Sturrock, P. A. (1971) Astrophys. J. 164, 529-556. holes, exploding galactic nuclei, and the primordial fireball 3. Ruderman, M. A. & Sutherland, P. G. (1975) Astrophys. J. were all mere theoretical constructs little more than a dec- 196,51-52. ade ago, there are now firm grounds for accepting them as 4. Kerr, R. P. (1963) Phys. Rev. Lett. 11, 237-239. real. Either we can legitimately extrapolateknown physical 5. Chandrasekhar, S. (1935) Mon. Not. R. Astron. Soc. 95, 207- laws to analyze these phenomena meaningfully, or else- 225, and earlier work cited therein. in a sense still more still 6. Schreier, E., Gursky, H., Kellog, E., Tananbaum, H. & Giac- which might be interesting-some coni, R. (1971) Astrophys. Lett. 170, L.21. unenvisioned laws of nature intervene to prevent this. 7. Blandford, R. D. & Rees, M. J. (1974) Mon. Not. R. Astron. If there has been any discernible general trend in high en- Soc. 169, 395-415, and references cited therein. ergy astrophysics over the last 10 years, it is that the tradi- 8. Longair, M. S., ed. (1974) "Confrontation of cosmological tional sub-disciplinary boundaries within the field have be- theory with observation," Proc. IAU Symposium 63 (Reidel, come blurred in several respects. It has become increasingly Holland) Contains reviews of most aspects of observational important to correlate observations of the same object in dif- cosmology. ferent wavebands. Similar physical processes may operate on 9. Penzias, A. A. & Wilson, R. W. (1965) Astrophys. J. 142, different scales: for example, the plasma physics involved in 419-421. neutron star 10. Robson, E. I., Vickers, D. G., Huizinga, J. S., Beckman, J. E. & the Earth's magnetosphere, magnetospheres, Clegg, P. E. (1974) Nature 251, 591-592. and extragalactic radio sources has many points of similari- 11. Woody, D. P., Mather, J. C., Nishioka, N. S. & Richards, P. L. ty; and superdense matter occurs both in neutron stars and (1975) Phys. Rev. Lett. 34, 1035-1037. in discussions of the very early universe. The character of 12. Hawking, S. W. (1971) Mon. Not. R. Astron. Soc. 152,75-78. the theoretical and interpretative problems is such that the 13. Hawking, S. W. Nature 248, 30-31. combined efforts of physicists spanning an increasingly 14. "Astronomy and Astrophysics for the 1970s," (Nat. Acad. Sci. broad range of expertise is required for their solution. USA, 1972). Downloaded by guest on September 27, 2021