New Perspectives

in Astrophysical
Cosmology
Second edition

Martin Rees
Kings College, University of Cambridge

PU BLI S H E D BY T H E P RE S S S YND I C AT E O F TH E UN I VE R S I T Y O F C AM BR I DG E

The Pitt Building, Trumpington Street, Cambridge, United Kingdom

CAM BR I DGE UN I VE RS I TY P RE S S

The Edinburgh Building, Cambridge CB2 2RU, uk http://www.cup.cam.ac.uk

40 West 20th Street, New York, NY 10011-4211, USA http://www.cup.org
10 Stamford Road, Oakleigh, Melbourne 3166, Australia
Ruiz de Alarco
n 13, 28014 Madrid, Spain
Cambridge University Press 1995, 2000
This book is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without
the written permission of Cambridge University Press.
First published 1995
Second edition 2000
Printed in the United Kngdom at the University Press, Cambridge
Typeset in Hollander 10/15pt [V N]
A catalogue record for this book is available from the British Library
Library of Congress Cataloguing in Publication data
Rees, Martin J., 1942
Perspectives in astrophysical cosmology / Martin Rees. 2nd ed.
p.

cm.

Includes bibliographical references and index.

ISBN 0 521 64238 8 (hc.)
1. Cosmology. 2. Galaxies. 3. Nuclear astrophysics. I. Title.
QB981.R37 2000
523.1dc21 9921389 CIP
ISBN 0 521 64238 8 hardback

Contents
Preface vii

1 The cosmological framework 1

Introduction 1
Large-scale structure: how homogeneous is the
universe? 2
High-redshift objects 5
Pre-galactic history 11
Status of the hot-big-bang hypothesis 15

2 Galaxies and dark matter 20

What are galaxies? 20
What is special about galactic dimensions? 23
Dark matter 26
What can the dark matter be? 30
How to discriminate among dark-matter options 38

3 Emergence of cosmic structure 41

Gravitational instability 41
The fluctuation spectrum at trec 45
Is the universe flat? 52
Classical methods for determining ) 57
v

CONTENTS

Clues from the microwave background 62

Dissipative effects for the baryon component 67
Is any simple hypothesis compatible with all
the data? 68

4 Quasars and their demography 74

Quasars and the epoch of galaxy formation 74
How many quasars have there been? 81
Quasar masses and efficiencies 82
Dead quasars: massive black holes in nearby
galaxies 84
Binary black holes? 93
Cosmogonic interpretations of quasar evolution
some speculations 95

5 Some probes and relics of the high-redshift

universe

100
Quasars as probes of intervening gas 100
The epoch z 9 5 105
Magnetic fields 112
Cosmic strings 118

6 Some fundamental questions 123

Gravity 123
The ultra-early universe 125
Flatness and the horizon problem 128
Inflationary models 129
Concluding homily 136
References 141
Some further reading 147
Author index 149
Subject index 153

The cosmological framework

Introduction
Gravity, almost undetectable between laboratory-scale bodies, is
the dominant force in astronomy and cosmology. The basic
structures in our cosmic environment stars, galaxies, and
clusters of galaxies all involve a balance between gravitational
attraction and the disruptive effect of pressure or kinetic
energy. Our entire observable universe may display a similar
balance: the Hubble expansion is being slowed (and may perhaps eventually be braked to a halt) by the gravitational effect of
its entire mass-energy.
The best-understood cosmic structures are the smaller ones:
the individual stars. Stellar structures and life-cycles can be
predicted theoretically, and tested empirically by observing
large populations of stars, of differing ages, in the Milky Way.
The Milky Way, the disc galaxy to which the Sun belongs, can be
envisaged as a kind of ecological system in which stars are
continually being born and dying, their gaseous content being
recycled and chemically enriched as the evolution proceeds.
Our own Galaxy is typical of the galaxies distributed through
1

TH E CO SM O LO GI CAL F RAM EWO R K

the universe, which are the most conspicuous features of the

cosmic scene. Why should the universe be full of these remarkable aggregates of stars and gas, typically ~ 105 light-years across
and containing around 1011 stars? We do not yet have compelling physical reasons for the characteristic properties of
galaxies, as we do for stars.
One reason why galaxies are harder to understand than stars
is that their formation impinges on cosmology. Individual stars
form, evolve, and die more or less regardless of what the
universe does initial cosmic conditions have left no traces on
the complex gas dynamics that goes on within each galaxy. But
that is not true for galaxies, which may have emerged, at an
epoch when the entire universe was denser and perhaps very
different, from inhomogeneities that were imprinted on the
universe in its earliest phases.

Large-scale structure: how homogeneous is the

universe?
In the perspective of the cosmologist, even entire galaxies are
little more than points of light which indicate how the material content of the universe is distributed, and how it moves.
Galaxies are clustered: some in small groups (like our own Local
Group, of which the Milky Way and the Andromeda galaxy are
the dominant members), others in big clusters with hundreds
of members. Moreover the clusters themselves are grouped in
filamentary or sheet-like superclusters. In recent years, there
has been great progress in quantifying the distribution of
galaxies over the sky, and also in mapping out the threedimensional structure. The latter task has entailed determining
redshifts and distances for thousands of galaxies.
Figure 1 shows the major groupings of galaxies within our
2

LARGE-SCALE STRUCTURE: HOW HOMOGENEOUS IS THE UNIVERSE?

Figure 1

The most conspicuous clusters and superclusters within a cube, of

dimensions around 3 ; 108 light-years (108 pc), centred on our own Galaxy.
There are also, of course, many galaxies more uniformly distributed in the
space between these clusters. The linear dimensions of the region depicted
here are about 2 per cent of the size of the part of the universe accessible to
optical observations. This cube is probably large enough to provide a fair
sample of the contents of the universe: on larger scales the amplitude of
inhomogeneities is much less than unity. (From Hudson, M. J. 1993, Mon. Not.
Roy. Astron. Soc. 265, 43 (Fig. 10).)

local part of the universe, out to a distance of around 3 ; 108

light-years. At still greater distances, galaxies are distributed
more uniformly over the sky. There is no evidence that big
density contrasts extend to larger scales. A region of the size
3

TH E CO SM O LO GI CAL F RAM EWO R K

shown in this figure is therefore probably large enough to

provide a fair sample of the contents of the universe.
Our universe certainly does not have a simple fractal structure, with clusters of clusters of clusters ad infinitum. There is a
definite upper limit to the scale on which large-amplitude
density inhomogeneities are observed. The largest structures
with (do/o) ) 1 are about 1 per cent of the Hubble radius: the
typical nonlinear scale is around 0.3 per cent. The typical metric
fluctuations due to clusters and superclusters defined in
dimensionless form as the gravitational energy per unit mass
arising from the associated density enhancement, in units of c2
have an amplitude of the order of Q = 105. The velocities relative to
the Hubble flow induced by these structures are typically below
Q1/2c = 1000 km s1. The mass-equivalent of the kinetic energy
associated with these so-called peculiar motions is therefore
only 105 of the rest mass. This number Q = 105, a measure of the
metric fluctuations in our universe, has an importance which
will come up again later. Its smallness implies that the present
local dynamics of cosmic inhomogeneities such as clusters and
superclusters can be validly approximated by Newtonian gravity. Even more importantly, it justifies the relevance of the
simple theoretical models for a homogeneous isotropic universe. These models date back to the 1920s.
The first relativistic models for a homogeneous expanding
universe were found by Friedmann before Hubble discovered
the recession of the nebulae. Hubbles work, which showed that
the universe did not resemble Einsteins earlier static model,
stimulated further studies of relativistic cosmology by
Lemaitre, Tolman, and others. But the data were then and
remained for several decades too sparse to indicate whether
any of these idealised models fitted the real universe, still less to
discriminate among them.
4

HI GH -R EDS HI F T O BJ E CT S

High-redshift objects
Hubbles work suggested that the galaxies would have been
crowded together in the past, and emerged from some kind of
beginning. But he had no direct evidence for cosmic evolution:
indeed the steady-state theory, proposed in 1948 as a tenable
alternative to the big bang, envisaged continuous creation of
new matter and new galaxies, so that despite the expansion the
overall cosmic scene never changed.
To discern any cosmic evolutionary trend, one must probe
objects so far away that their light set out when the universe was
significantly younger. This entails studying objects billions of
light-years away with substantial redshifts. A programme to
measure the cosmic deceleration was pursued from the 1950s
onwards with the 200-inch Palomar telescope. But the results
were inconclusive, partly because normal galaxies are not
luminous enough to be detectable at sufficiently large redshifts.
It was Ryle and his colleagues from radio astronomy, in the late
1950s, who found the first real evidence that the universe was
indeed evolving. Radio telescopes could pick up emission from
some unusual active galaxies (which are now believed to be
harbouring massive black holes in their centres) even when they
were too far away to be seen with optical telescopes. One cannot
determine the redshift or distance of such sources from radio
measurements alone, but Ryle assumed that, statistically at
least, the ones appearing faint were more distant than those
appearing intense. He counted the numbers with various apparent intensities, and found that there were too many apparently
faint ones in other words, sources at large distances compared with the number of brighter and closer ones. This was
discomforting to the steady statesmen, but compatible with an
evolving universe if galaxies were more prone to undergo
5

TH E CO SM O LO GI CAL F RAM EWO R K

violent outbursts in the remote past, when they were young.

The subsequent discovery by optical astronomers of extreme
active galactic nuclei (quasars) at very large redshifts corroborated Ryles conjectures, but these objects, and their evolution,
are still too poorly understood to be used for determining the
geometry of the universe.
By probing deep into space, astronomers can study parts of
the universe whose light set out a long time ago. If we lived in a
wildly inhomogeneous universe, there would be no reason why
these remote regions (and the way they have evolved) should
bear any resemblance to our own locality. However, insofar as
the universe we find ourself living in (or at least the part of it
accessible to observation) is actually uniform and isotropic, its
gross kinematics are describable by a single scale factor R(t); all
parts have evolved the same way and have the same history (see
Figure 2). This simplicity gives us reason to believe that when we
observe a region of the universe that lies (say) 3 billion lightyears away, its gross features (the statistical properties of the
galaxies, the nature of the clustering, etc.) resemble those that
would have been displayed 3 billion years ago in our own
locality (i.e. within the region depicted in Figure 1).
Astronomers have an advantage over geologists, in that they
can directly observe the past. And there has been spectacular
progress in the technology for probing faint and distant objects. The first improvement came when photographic plates
were replaced by CCD solid-state detectors up to 50 times more
sensitive at optical and near infrared wavelengths. The advent of
a new generation of telescopes with 10-metre mirror diameters
has enhanced astronomers abilities to study the light from
faint objects. (The two Keck Telescopes in Hawaii are already
complete; several more are currently being built.)
The faintest and most distant galaxies appear typically only
6

HI GH -R EDS HI F T O BJ E CT S

Figure 2

Schematic space-time diagram showing world line of our Galaxy and our past
light cone. The only regions of space-time concerning which we have direct
evidence are those shaded in the diagram, which lie either close to our own
world line (inferences on the chemical and dynamical history of our Galaxy,
geological evidence, etc.) or along our past light cone (astronomical
evidence). It is only because of the overall homogeneity that we can
confidently assume any resemblance between the distant galaxies whose
light is now reaching us and the early history of our Galaxy. In homogeneous
universes we can define a natural time coordinate, such that all parts of the
universe are similar on hypersurfaces corresponding to a given value of t.

13 arcseconds across, and are little more than blurred

smudges of light when viewed from the ground, because atmospheric fluctuations smear even a point source over a substantial
fraction of an arcsecond. But the Hubble Space Telescope, after
its optics was corrected in 1994, has yielded much sharper
pictures. The most spectacular single image, the so-called
Hubble Deep Field, was obtained by pointing the telescope for
more than a week towards the same patch of sky.5a Observations
7

TH E CO SM O LO GI CAL F RAM EWO R K

with this level of sensitivity reveal several hundred galaxies,

with a range of morphologies, within a patch only an arcminute
square. Redshifts have been measured for many of these, using
the Keck Telescope.5b In many cases the wavelengths are
stretched, between emission and reception, by a factor Rnow/
) is
Rem = 1 + z 9 4: the absorption edge at the Lyman limit (912 A
shifted into the visible band, and is indeed the most prominent
feature in the spectrum. Larger samples of high-redshift galaxies have been discovered by using this distinctive spectral
feature shifted into the blue part of the visible spectrum as a
diagnostic.5c
The light from these remote galaxies set out when the
universe was much younger than it is today: we are observing
them at a stage when they are only recently formed, and it is not
surprising to find that they look distinctively different from
nearby systems.
There have been astonishing advances, during the late 1990s,
in detecting galaxies at very high redshifts. The observation of
high-redshift objects is, however, not in itself so novel: quasars
and other active galactic nuclei (e.g. the intense radio sources),
the hyperactive centres of a special subset of galaxies, outshine
the stellar content of their host galaxy by a factor that can
amount to many thousands. These are so bright that highquality spectra could be taken even with moderate-sized telescopes. An early example of a high-redshift quasar is PC
1247 + 3406, with z = 4.89, whose spectrum is shown in Figure 3;
line is observed in the red part of the
the Lyman-a 1216 A
. To estimate the relative age of the
spectrum, at around 7200 A
universe then and now, one needs to know the dynamics of the
expansion, and in particular how much it has been decelerating. If there were no deceleration at all, the universe would have
been younger when the light set out by the factor 1 + z of 5.89.
8

HI GH -R EDS HI F T O BJ E CT S

Figure 3

The spectrum of the quasar PC 1247 + 3406, with redshift z = 4.89. Light from
this object set out towards us when the cosmic scale factor R was
(1 + z) = 5.89 times smaller than it is today. According to the Einsteinde Sitter
model, the universe would then have been only ~ 7 per cent of its present age.
(From Schneider, D. P., Schmidt, M. & Gunn, J. E. 1991, Astron. J. 102, 837.)

However, according to the Friedmann models the expansion is

decelerating. In the theoretically attractive Einsteinde Sitter
cosmology the scale factor of the universe grows as R . t2/3. The
light now reaching us from PC 1247 + 3406, according to that
model, would have set out when the universe was younger by a
factor 5.893/2. Astronomers can therefore probe the last 90 per
cent of cosmic history. The existence of these quasars tells us
that by the time the universe was about 109 years old some
galaxies (or at least their inner regions) had already formed, and
9

TH E CO SM O LO GI CAL F RAM EWO R K

runaway events in their centres had led to the extreme type of

active nuclei that the quasar phenomenon represents.
The host galaxies of quasars should presumably have formed
before the quasars themselves; moreover, if galaxies build up
hierarchically, smaller galaxies (perhaps themselves too small
to host a powerful quasar) should form still earlier. There is
therefore every reason to expect galaxies with redshifts substantially larger than 5. These would generally be very faint
certainly too faint for a high-quality spectrum to be obtainable
even with a 10 m telescope. However, some faint fuzzy objects
with z 9 5 have been found by extending to higher redshifts the
techniques that proved so successful in finding galaxies with
z = 3.6a Another technique for finding them is to use filters for
objects whose low-resolution red spectra exhibit a line that is
actually highly redshifted Lyman a.6b Such attempts have already revealed several galaxies further away than PC 1247 + 3406:
in one or two cases the effort is aided by the lucky accident that
those galazies are gravitationally lensed (see Chapter 2) by a
cluster of galaxies along the line of sight.6c It is not clear what
the limiting galaxy redshift will be: it depends on how and
when galaxy formation starts, a topic discussed further in
Chapter 5.
The light from bright quasars offers an important probe for
the intervening medium. Absorption lines blueward of Lyman a
in the spectrum indicate clouds of gas lying along the line of
sight. The absorption is probably caused by protogalaxies too
faint to be seen by their direct emission (and where perhaps no
stars have yet formed). The way this absorption depends on
redshift offers important clues to how galaxy formation proceeded; this is discussed further in Chapter 5.

10

PRE - GA L ACT I C H IS T OR Y

Pre-galactic history
But what about still earlier epochs, before any galaxies could
have formed? Did everything really emerge from a dense (or
even singular) beginning ten or fifteen billion years ago? The
clinching evidence dates back to 1965, when Penzias and Wilson published their classic paper announcing excess antenna
temperature at 4080 Mc/s. Intergalactic space is not completely
cold but has a temperature of about 3 K. This may not sound
much, but it implies that there are about 4 ; 108 photons per
cubic metre maybe a billion photons for every atom in the
universe.
The discovery of the microwave background quickly led to a
general acceptance of the so-called hot-big-bang cosmology a
shift in the consensus among cosmologists as sudden and
drastic as the shift of geophysical opinion in favour of continental drift that took place at about the same time. There seemed
no plausible way of accounting for the microwave background
radiation except on the hypothesis that it was a relic of an epoch
when the entire universe was hot, dense, and opaque. Moreover,
the high intrinsic isotropy of the radiation meant that the
simple mathematical models were a better approximation to
the real universe than the theorists who devised them in the
1920s and 1930s would have dared to hope. Subsequent
measurements of this background, made with increasing precision at various wavelengths, have strengthened these conclusions. The radiation spectrum is now known, primarily
through the magnificent results from Mather and his collaborators, using the FIRAS (Far Infrared Absolute Spectrophotometer) experiment on the COBE (Cosmic Background Explorer)
satellite, to deviate from a black body by less than 1 part
in 104. The best-fitting temperature is 2.726 K. And measure11

TH E CO SM O LO GI CAL F RAM EWO R K

ments by several groups9 show that the radiation is intrinsically isotropic to within a few parts in 105, but that there are
apparent anisotropies, on angular scales from 0.3 up to 90, at
the 105 level (some quantitative implications of these are
mentioned in Chapter 3).
In the dense early phases, the radiation would have been held
in thermal equilibrium with the matter scattering repeatedly
off free electrons whose density would have been high enough
to make the universe very opaque. But when expansion had
cooled the matter below 3000 K (when the cosmic scale factor R
was 103Rnow) the primordial plasma would have recombined,
leaving few free electrons. The fog would then have lifted, the
universe thereafter becoming transparent, and probably remaining so until the present (see p. 108). The experimentally
detected microwave photons are direct messengers from an era
when the universe was about a thousand times more compressed, and had expanded for about half a million years. But
the photons are still around they fill the universe and have
nowhere else to go. An important cosmic number is the
photon-to-baryon ratio g1, which stays essentially constant
during the cosmic expansion. It is because this ratio is large that
many authors refer to the hot big bang.
The universe contains other important fossils of a cosmic era
far earlier than (re)combination: the light elements such as D,
3
He, 4He, and 7Li. During the first minute of cosmic expansion,
when temperatures were above 109 K, nuclear reactions would
have synthesised these elements, in calculable proportions,
from protons and neutrons. The baryon density in an expanding universe goes as R3 . T 3, and would therefore have been
1027 times higher when T = 3 ; 109 K than it is today. But this is
still not as high as the density of air! One does not need to worry
about problems of dense matter. And the energies of the
12

PRE - GA L ACT I C H IS T OR Y

relevant nuclear reactions are :1 MeV, and do not involve any

large uncertain extrapolation from the experimental domain.
Such calculations, showing how the light-element abundances would depend on the present mean baryon density, the
number of neutrino species, etc. were done in the 1960s.
Although refinements have been introduced, nothing essential has changed on the theoretical front over the last 25 years.
Stellar nucleogenesis, supernova explosions, and recycling
into new stars, the theory of which was formulated in the
1950s, seem well able to account for heavy elements such as
carbon, oxygen, and iron. But the high and relatively uniform
proportion of helium always posed a problem. It was therefore
gratifying, and neatly complementary, that helium was the one
element that would be created prolifically in a big bang. In the
1970s, the astrophysical problems of accounting for deuterium
(whose abundance is reduced during stellar recycling) were
properly appreciated, and this isotope is also believed to be a
cosmological fossil.
Only more recently have astronomers been able to determine
the light-element abundances in old stars, gaseous nebulae, etc.
precisely enough to make a worthwhile comparison with the
big-bang predictions. In particular, the helium abundance is
now pinned down with a precision approaching 1 per cent.
Measurements of deuterium in our own Galaxy yield a lower
limit to the primordial abundance, because an uncertain proportion would have been destroyed by processing through
earlier generations of stars. It was therefore an important
advance when the Keck Telescope allowed astronomers to take
such high-quality spectra of quasars that weak lines due to D
(displaced from the much stronger H-lines by an isotopic shift
equivalent to 80 km s1) could be measured.17a These observations, referring to diffuse gas at early epochs, are likely to give a
13

TH E CO SM O LO GI CAL F RAM EWO R K

Figure 4

The predicted abundances of the light elements emerging from a standard

hot big bang, as a function of the baryon/photon ratio g. Note that there is a
definite range of g for which the calculations yield abundances of 4He, D, 3He,
and 7Li concordant with observations. (From Schramm, D. N. 1991, in After
the First 3 Minutes, eds. Holt, S. S. et al. (American Institute of Physics, New
York) p. 12.)

better estimate than local measurements of the actual primodrial abundance of deuterium.
What is remarkable is that, as Figure 4 shows, the lightelement abundances all appear concordant with the predictions of big-bang nucleosynthesis, provided that the baryon
density is in the range 0.1 to 0.3 baryons per cubic metre (a
density compatible with what we observe). The measured abundances could have been all over the place, or could have
indicated a mean cosmic density that was plainly ruled out;
these nucleosynthesis calculations therefore offer a strong
vindication for extrapolating a standard big-bang model back
14

ST AT US O F T HE HO T -B I G- B A N G HY PO T HE SI S

to a temperature T such that kT = 1 MeV. The grounds for this

extrapolation should, I believe, be taken as seriously as, for
instance, ideas about the early history of our Earth, which are
often based on indirect inferences by geologists and paleontologists that are a good deal less quantitative.

Status of the hot-big-bang hypothesis

I would bet odds of 10 to 1 in favour of the general hot-big-bang
concept as a description of how our universe has evolved since it
was around 1 second old and at a temperature of 1010 K (or ~ 1
MeV). Some people are even more confident. In a memorable
lecture at the International Astronomical Union back in 1982,
Zeldovich claimed that the big bang was as certain as that the
Earth goes round the Sun. He must even then have known his
compatriot Landaus dictum that cosmologists are often in
error but never in doubt!
The case for the standard hot big bang has actually
strengthened greatly in the last decade, through better
measurements of the background radiation and of the light
elements. Moreover, one can think of several discoveries that
could have refuted the model and which have not been made. For
instance:
(i) Astronomers might have discovered an object whose helium abundance was zero, or at any rate well below 23 per
cent. (Stellar nucleosynthesis can readily enhance helium
above its pre-galactic abundance, but there seems no astrophysically plausible way of eliminating it.)
(ii) The background radiation spectrum might, as experimental precision improved, have turned out to be embarrassingly different from a black body. In particular, the mil15

TH E CO SM O LO GI CAL F RAM EWO R K

limetre-wave background measured by COBE might have

been below a black-body extrapolation of what had already
been reliably determined at centimetre wavelengths. It
would not be hard to think of effects that would have added
extra radiation at millimetre wavelengths (indeed the
smallness of the millimetre excess strongly constrains the
input from early star formation, decaying particles, etc.),
but it would be hard to interpret a millimetre-wave temperature that was lower than a black body fitting the
RayleighJeans part of the spectrum.
(iii) If a stable neutrino had been discovered in the mass range
from 100 to 106 eV, that would have been incompatible with
the standard big-bang model, which would predict about
1.1 ; 108 such neutrinos per cubic metre; relic neutrinos
would then provide a far higher density in the present
universe than is compatible with observations.
These considerations give us confidence in extrapolating
right back to the first few seconds of our universes history and
in assuming that the laws of microphysics were the same then
as now. Conceivably, this confidence is misplaced, and our
satisfaction will prove as transitory as that of a Ptolemaic
astronomer who has fitted a new epicycle. But the hot big bang
certainly seems vastly more plausible than any equally specific
alternative.
If we envisage time on a logarithmic scale, then many
important events of cosmic history are being overlooked if we
consider only the period t 9 1 s. Figure 5 depicts how the
universe might have evolved right from the Planck time to the
present. Uncertainties about the relevant physics impede our
confidence in discussing the extensive span of logarithmic time
1043 s : t : 104 s, when thermal energies exceeded 100 MeV. At
16

ST AT US O F T HE HO T -B I G- B A N G HY PO T HE SI S

Figure 5

The history of our universe according to the standard hot-big-bang model,

showing some of the key physical processes at various stages.

later times, we can consistently use microphysics that is well

known; so long as the universe remains almost homogeneous,
the evolution is straightforwardly calculable. However, at some
stage small initial perturbations must have evolved into gravitationally bound systems (protogalaxies? protoclusters?); the
onset of nonlinearity then creates challenging complications,
even though the controlling physics is Newtonian gravity and
gas dynamics.
Despite the importance and fascination of the ultra-early
universe it would be imprudent to venture any bets on what
happened when t  1 s. The empirical basis for these initial
17

TH E CO SM O LO GI CAL F RAM EWO R K

phases of cosmic history is far more tenuous than the quantitative fossil evidence (from light elements and the background
radiation) for the eras after 1 second. The first millisecond of
cosmic history, a brief but eventful era spanning 40 decades of
logarithmic time (starting at the Planck time), is the intellectual
habitat of the high-energy theorist and the inflationary or
quantum cosmologist. Densities and energies were then so high
that the relevant physics is speculative.
From 103 seconds onwards, quantitative predictions, such as
those about cosmic light-element production, are possible;
these vindicate our backward extrapolation. (These predictions
also, incidentally, vindicate the assumption that the laws of
microphysics were indeed the same when the universe had been
expanding for only 1 second as they are in our terrestrial
laboratories; we should keep our minds open or at least ajar
to the possibility that this isnt so.)
There has been remarkable progress in the last 25 years in
delineating cosmic evolution, mapping out the structure and
dynamics of clusters and superclusters, and surveying objects at
high redshifts. This progress brings new, strongly interrelated,
questions into sharper focus:
(i) How did the dominant present-day structures in our universe galaxies and clusters emerge from amorphous
beginnings in the early universe?
(ii) What is the dark matter that seems to be the dominant
constituent of the universe?
(iii) Are the key parameters that have determined the nature of
the present-day universe the structure, the baryon content, the dark matter, etc. a legacy of exotic physics in the
ultra-early phases?
These lectures are mainly concerned with the first two of
18

ST AT US O F T HE HO T -B I G- B A N G HY PO T HE SI S

these questions (which are interlinked), but the final chapter

will touch briefly on current conjectures that relate more
directly to the still-mysterious physics governing the earliest
phases of cosmic history.

19