91É«ÇéƬ

At the beginning of the year, a team of British astronomers published
a research paper in the journal Nature that, according to the headline writers
of the world’s newspapers, plunged cosmology into crisis. Our paper described
the results of a systematic study, using ground-based telescopes, of galaxies
first found at infrared wavelengths by the Infrared Astronomical Satellite
(IRAS). We had made the first detailed reliable three-dimensional maps of
the distribution of matter in the Universe. The results were dramatic: they
revealed a serious discrepancy in the current standard model of how the
Universe formed and evolved, a model that relies on an exotic, invisible
form of matter -‘cold dark matter’ – to explain how galaxies and clusters
of galaxies formed.

For 10 years, we have known that there is more to the Universe than
meets the eye. Astronomers have come to the conclusion that the Universe
cannot only contain the normal matter that our bodies, the Earth and the
stars are made of. The bulk of this normal matter, the matter that we see
through our telescopes, is in the form of baryons – that is, protons and
neutrons. Other particles, such as electrons, contribute little to the mass
of the Universe.

Not long after the big bang – the fireball in which our Universe was
created – baryons somehow clumped together to make galaxies and clusters
of galaxies. But at the very time this process should have been well under
way, we find that the Universe was, in fact, incredibly smooth and uniform.
We know this from observations of the microwave background radiation, which
is believed to be the faint relic of the fireball phase of the big bang.
The remarkable smoothness of this radiation shows that, some 300 000 years
after the big bang, the normal, baryonic matter and radiation were uniform
to at least one part in 10 000. From such a smooth state, there is simply
not time for gravity to have assembled the galaxies and clusters we see
today. The explanation astronomers favour is that the Universe must also
contain matter in some, so far undetectable dark form.

Theorists came up with two contrasting possibilities for this dark matter.
One was ‘hot’ dark matter consisting of particles moving at speeds close
to that of light. The other variety of exotic particle allowed by theory
emerges from the big bang with low velocities. This is called ‘cold’ dark
matter. In the 1970s, when theorists first aired the idea of dark matter,
there was an immediate candidate for the hot variety in the form of neutrinos.
Particle physicists usually regard neutrinos as having no mass and moving
at precisely the speed of light. The possibility, however, that they have
a small mass has not been ruled out. There are no known particles with the
right properties to be cold dark matter, but particle physicists predict
several possible candidates. The two most popular particles, which fit in
with ideas of how the forces of nature are unified, are known as the axion
and the photino.

In the very early Universe, all forms of matter and radiation interact
strongly with each other and share their energy equally. As time goes by,
different components of matter separate out from the radiation. They then
start to condense together under the action of gravity. The weakly interacting
particles that make up the dark matter would have ‘decoupled’ from the radiation
at a much earlier time than the ordinary baryonic matter, and any irregularities
in density would have had plenty of time to grow more pronounced under the
action of gravity. By the time the baryonic matter decoupled from the radiation,
the dark matter irregularities would already be well developed and would
act as seeds for the condensation of the baryonic matter.

Hot and cold dark matter do this in quite different ways. Calculations
show that hot dark matter would have formed structures on very large scales.
These would collapse to give pancake-shaped aggregates out of which galaxies
form. This is the ‘top-down’ picture for the formation of galaxies and clusters.
Cold dark matter, on the other hand, would first form clumps on small scales,
which act as seeds for galaxy formation. Galaxies would then aggregate into
clusters and larger structures under the action of gravity. This is the
‘ bottom-up’ scenario.

Because it was difficult to match the top-down picture to the actual
distribution of galaxies in clusters, it was the cold dark matter scenario
that was pursued more vigorously. Computer simulations of the way galaxies
form and cluster were explored by a group of astronomers – Marc Davis from
the University of California at Berkeley, George Efstathiou from the University
of Cambridge, Carlos Frenk from the University of Durham and Simon White
from the University of Arizona. Even with the aid of cold dark matter, they
had to assume that galaxy formation was ‘biased’, so that galaxies form
only where there are exceptionally large fluctuations in the density of
dark matter. Otherwise, the simulations would have produced a distribution
of galaxies that was far too bland and featureless.

Not only could the cold dark matter scenario accurately predict the
average probability of clustering of galaxies, but the computer simulations
also looked very similar to the distribution of galaxies we see in the sky.
But was there any evidence that the Universe actually contained dark matter?

Studies with radio and optical telescopes of the way in which spiral
galaxies rotate suggested that they are surrounded by haloes of dark material.
Analysing the motions of pairs and groups of galaxies confirmed this result:
galaxies are heavier than the total amount of visible, radiating matter
would suggest. But when the masses of all galaxies, including their dark
haloes, are added together to give an average density for the Universe,
the total comes to no more than about 10 per cent of what is called the
‘critical’ value. All universes with less than the critical density will
expand for ever, while in one of higher density, gravity will ultimately
halt the expansion and the universe will collapse again to a ‘big crunch’.

Theorists tend to favour the Universe having almost exactly the critical
density, partly because it avoids the coincidence of us existing at a rather
special epoch – when the kinetic energy of the receding galaxies almost
balances the gravitational energy. But the main reason cosmologists prefer
the Universe to have a critical density is that it is a requirement of a
particularly successful theory of the early Universe called ‘inflation’.
Alan Guth, of the Massachusetts Institute of Technology, proposed inflation,
in 1981, as a phase that happened just after the big bang, when the whole
of the observable Universe inflated exponentially from an infinitesimally
small volume.

We can also deduce the density of the Universe in baryons from the relative
abundances of the light elements, helium, deuterium and lithium. These elements
were made during the fireball phase of the big bang and the amount produced
depends sensitively on the density of the Universe. The result of this calculation
is that the average density of the Universe in the form of baryons is about
5 per cent of the critical value. But because inflation predicts that the
Universe should still be very close to the critical density, it requires
that 95 per cent of the matter in the Universe today should be in some dark,
non-baryonic form.

So dark matter is not only needed to explain how galaxies condensed
from the highly smooth state of the Universe before matter and radiation
decoupled, but is also predicted by inflationary cosmology. The cold dark
matter scenario ties together particle physics, cosmology and the theories
of galaxy formation and clustering into an elegant package. This is why
particle physicists and astronomers think so highly of the theory.

Why is the cold dark matter scenario in difficulty today? No matter
how impressive a cosmological theory is, it has to fit in with what we actually
see in the sky. Let us examine first some of the observations of the galaxy
distribution made over the past decade. In 1983 Marc Davis, John Huchra
and colleagues at Harvard measured the distances of more than 2000 galaxies
by observing their red shifts with optical telescopes on Earth. The light
from galaxies is shifted towards the red end of the visible spectrum (to
longer wavelengths) because the galaxies are moving away from us. In an
expanding Universe, the velocity at which they are moving away is proportional
to the distance of the galaxy.

Huchra and his colleagues were able to map the concentration of galaxies
centred on the Virgo cluster, known as the Local Supercluster, and to explore
galaxy clustering as far out as 30 million light years. More recently, surveys
deeper in space by Margaret Geller and Huchra revealed the existence of
the ‘Great Wall’, a sheet of galaxies 300 million light years long joining
the Coma and Hercules superclusters. They also confirmed the existence of
substantial voids almost free of galaxies. The first such void to be identified
was the Bootes void, found in a survey of a small area of the sky by Robert
Kirschner, also at Harvard, and others. These walls and voids posed some
difficulties for the cold dark matter theory, though they could for the
moment be dismissed as anecdotal evidence, statistical anomalies. The coup
de grace was delivered by the survey of red shifts of galaxies detected
by IRAS which I and my colleagues carried out.

In 1976, the US, the Netherlands and Britain decided to collaborate
in sending up a cryogenically cooled telescope, IRAS, to survey the skies
at far infrared wavelengths (10 to 100 micrometres). I was convinced that
we could use it for cosmological studies. When the satellite was launched
in February, 1983, I was temporarily installed at NASA’s Jet Propulsion
Laboratory in Pasadena, leading the group responsible for the computer programs
which were to analyse the data for individual sources of infrared radiation.

When IRAS was originally designed, we had hoped to see hundreds of thousands
of sources including tens of thousands of galaxies. The performance of the
detectors in tests on the ground, however, had been so poor that I thought
we would see only 1000 sources, of which very few were expected to lie outside
our Galaxy. To our amazement, once in orbit, the telescope and detectors
performed so perfectly that our original plans and hopes were easily exceeded.

Within a few weeks of launch, we had completed a thorough survey of
a strip of the sky. For months, we analysed this data, tuning the computer
algorithms used to detect and confirm the reality of the sources of infrared
radiation. In July 1983, we made the final adjustment to the hundreds of
parameters needed to catalogue the IRAS sources. The computer then started
to generate these catalogues. It was to take almost a year to complete,
with one day spent computing for each day’s worth of data. The IRAS mission
lasted 10 months before the liquid helium coolant ran out, by which time
it had surveyed 96 per cent of the sky.

By comparing the positions of the infrared sources with those of sources
in optical catalogues and on photographic plates, it soon became clear that
IRAS was picking up thousands of galaxies at the longer wavelengths of 60
and 100 micrometres. By the time we released the IRAS Point Source Catalog
in November 1984, I was already planning a programme to study these IRAS
galaxies from ground-based optical telescopes. I had persuaded Andy Lawrence
from the Royal Greenwich Observatory to join us at Queen Mary and Westfield
College in London, and we set up a collaboration with Michael Penston at
the Royal Greenwich Observatory, who died so tragically last December. With
Michael’s research student Kieron Leech, we measured the spectra of all
the IRAS sources in an area of 844 square degrees around the North Galactic
Pole. We showed that, apart from a few stars, obvious from their infrared
signature, more than 99 per cent of the sources at 60 micrometres were galaxies.

At this point, we conceived the idea of an ‘all-sky IRAS galaxy red
shift survey’, in which we would measure the red shifts, and thus distances,
of a large sample of 2400 IRAS galaxies spread all round the sky. We drew
in additional collaborators, Richard Ellis and Carlos Frenk from the University
of Durham, and George Efstathiou and Nick Kaiser from the University of
Cambridge, together with my two research students John Crawford and Will
Saunders. I was rather pleased with the acronym QCD (for QMW-Cambridge-Durham)
for the collaboration, but when Nick moved to the University of Toronto
and George to the University of Oxford, we had to change it to the less
stylish QDOT. We won the observing time for this survey in a gradual and
piecemeal way from the committees that allocate time on ground-based optical
telescopes.

In July 1987, when I organised the 3rd International IRAS Conference
at QMW, we had completed only 25 per cent of the survey. Marc Davis and
Michael Strauss at Berkeley, were already presenting preliminary results
from an all-sky survey of brighter IRAS sources. I was in despair. Then,
by a great stroke of luck, we successfully obtained the very first block
of observing time for research on Britain’s new 4.2-metre William Herschel
Telescope, a commissioning slot of four weeks in December 1987. We were
warned that the time might all have to be used for fixing the telescope.
In the event, the telescope worked almost perfectly.

The plan was to measure more than 50 red shifts a night, and we divided
up into roughly four teams of two, each working for a week. The first two
teams, George with Carlos and then Nick, were unlucky with the weather.
Richard and I also began with frustrating nights of bad weather. Then, when
12 of our 28 nights had passed with almost no red shifts measured, the weather
cleared and we set off at a maniacal rate. On our fifth night, Richard and
I measured 104 red shifts, almost certainly a world record, and something
like this pace was maintained to the end of the run, by which time we had
measured red shifts for 1100 IRAS galaxies and completed 80 per cent of
our survey. Two runs in the southern hemisphere on the Anglo-Australian
Telescope at Siding Springs in New South Wales allowed us to complete the
survey the following year. The long task of analysing the data began, much
of it carried out by Xia Xiaoyang of Beijing University, then studying for
a PhD at the University of Durham.

The Universe in 3D

Once we had the complete set of galaxy red shifts, we first studied
the distribution of the brightness of galaxies at infrared wavelengths.
We needed this data, both to study whether the infrared radiation emitted
from galaxies changed with cosmological epoch (it did), and to use the IRAS
galaxies to map out the total density of matter in the Universe. Our infrared
survey could map over a much larger fraction of the sky than any previous
optical survey. This is because visible light is absorbed by interstellar
dust which pervades our galaxy. Because IRAS was so sensitive we could also
probe further into space. We were able, therefore, to make a unique three-dimensional
map of the density of matter in a substantial volume of the local Universe,
of radius 500 million light years.

Using this density map, we could estimate the gravitational pull exerted
on our own Galaxy by the galaxies and clusters in our vicinity. The direction
of the net pull on us agreed well with the direction in which our Galaxy
is moving with respect to the microwave background radiation. We had, therefore,
fully explained the only unevenness observed in the otherwise perfectly
smooth microwave background – it is boosted (due to the Doppler effect)
in the direction in which we are moving, and depressed in the opposite direction.

Interestingly, our density map showed no sign of the so-called ‘ Great
Attractor’. Some astronomers had claimed that the Centaurus cluster was
‘streaming’ away from us at 1000 kilometres per second because it was being
pulled towards a large dense structure behind the cluster. Subsequent studies
of the Centaurus region have come to conflicting conclusions about this
streaming motion. Galaxy maps have shown that the clusters in this region
are very extended and that there is a new normal-sized cluster behind Centaurus,
but have not yet revealed a structure that could cause the streaming. Our
IRAS maps revealed that our Galaxy is being pulled by a dozen or so clusters,
with the bulk of the net pull being due to the Virgo, Hydra and Centaurus
clusters.

Saunders, as part of his PhD project, then went on to create three-dimensional
density maps of the Universe, based on the distribution of IRAS galaxies.
After completing his PhD and moving to Oxford, he transformed these to lurid
colour pictures (see Figure 1). These are really the first reliable, deep
density maps of the Universe available to astronomers. He also generated
a histogram of the density distribution (see Figure 2) which could then
be compared to the predictions of models, for example, the cold dark matter
model. Efstathiou had already devised a new statistical test of the models,
‘counts in cells’. The results, published in Monthly Notices of the Royal
Astronomical Society in December, were clearly inconsistent with the cold
dark matter theory. This paper, however, went almost unnoticed.

Saunders’ histograms, and the statistical test he calculated from them
showed the result much more clearly. The Universe is lumpier on the large
scale than predicted by the current version of the cold dark matter theory.
Our paper demonstrating this result was submitted to Nature, which decided
to print the colour density maps on its front cover. The journal also ran
a feature concluding that our paper meant the end of cold dark matter, and
put out a sensational press release. The result was that reports on the
paper appeared on the front pages of newspapers around the world, including
the New York Times and The Times of India.

Is this result really the end of cold dark matter? Our analysis of the
pull acting on our Galaxy using the data on IRAS galaxies indicates that
the Universe must have a density close to the critical value and, therefore,
is 90 per cent, or more, composed of some form of dark matter. We knew we
needed this dark matter to account for the formation of galaxies despite
the astonishingly smooth microwave background radiation. The issue is, therefore,
not whether there is dark matter in the Universe, but what form it takes.
Cold dark matter is very effective at explaining how galaxies and clusters
formed. But it seems to have a problem explaining the lumpiness we see on
large scales.

There are three choices. The first is to modify the cold dark matter
scenario to give more structure on large scales. It remains to be seen whether
theorists can do this convincingly. The second is to introduce some new
ingredient, for example, a component of hot dark matter, to account for
the large scale structure. Finally, the third choice is to try to come up
with some completely new explanation of galaxy formation and clustering.
Theorists are vigorously pursuing all three avenues.

On the observational front, a new consortium, QCCOD (QMW-Cambridge-Caltech-Oxford-Durham),
has embarked on a survey of the red shifts of IRAS galaxies to map out individual
structures even further away. The optical galaxy surveys carried out at
Harvard, giving fuller coverage of nearer objects across the sky, are also
continuing. The next few years will be exciting times for cosmology.

There were two sad ironies about the day that our paper was published
in Nature. In the morning, I attended a meeting of the British Space Science
Programme Board, at which we learned of the massive scale of the cuts that
the Government and the Science and Engineering Research Council have decided
to inflict on ground-based and space astronomy. So on the day we demonstrated
rather well the effectiveness of Britain’s role in both areas of astronomy,
the possibility of such a project being pursued in the next decade receded
from view. And then in the afternoon, Andy and I drove up to the funeral
of Michael Penston, who had been in with us at the beginning of the project.
He was a strong believer in the popularisation of astronomy and would have
enjoyed the wide coverage.

Michael Rowan-Robinson is a professor of astrophysics at Queen Mary
and Westfield College, London. His book Universe was published by Longman
last September.