One of my favourite astronomy memories dates from the mid-1990s. I had gone out to the Mauna Kea Observatory in Hawaii to measure the brightness of some high-redshift radio galaxies—a worthy but routine piece of science—and discovered that they were fainter than expected. This may not sound terribly interesting, but it was exciting for me and the other 100 or so scientists around the world who care about these objects. My memory, however, is of what happened after the observing run. During the night, we received an email that the space shuttle was due to pass over Hawaii. Just before dawn we went out of the telescope dome to watch. I had been expecting the shuttle to look something like a plane, but instead a bright point of light shot upwards at a crazy angle, crossing the sky in a couple of seconds, an upside-down meteor hurled towards the gods.
With time, this memory has come to stand for a big shift in my discipline. The space shuttle involved hundreds of collaborators and budgets of millions. As for my project, apart from the telescope operator, I was observing by myself, and the only other person involved was my friend Steve Rawlings, who was at home in Oxford. How things have changed in ten years. The global astronomy community has now just finished applying for observing time on the Herschel Space Observatory, a €1bn project due to be launched by the European Space Agency in early 2009. Last autumn, after finally submitting my own proposal, I counted the number of collaborators on it. There were 101. In ten years, astronomy has gone from a multitude of small-scale projects to Big Science schemes like Herschel.
One reason for this is the development of cameras and other instruments capable of observing large chunks of the sky in a single shot. It now makes sense to get together a large team who will carry out a large survey that can then be used for multiple projects. My own Herschel programme is a good example. Herschel will operate in the submillimetre waveband—the region of the electromagnetic spectrum between the infrared and radio wavebands—in which the universe is still largely unexplored, and so there is a good chance that our survey will discover something unexpected. But the data will also be used for many different projects, ranging from the study of protostars to galaxy evolution. Another reason for the change to this kind of big international project from the old-style small-scale projects is the revolution that has occurred in our understanding of the universe in the past ten years.
Two and a half facts
The Doppler shifts of the galaxies are Longair's half a fact, since they don't unambiguously imply there was a big bang. The two full facts are these. First, the universe is filled with an ocean of radio waves—the cosmic background radiation. This radiation is surprisingly strong (and is a small but appreciable addition to the noise on a television set). Second, the universe contains a very large amount of helium, the second lightest of the chemical elements, after hydrogen. Taken together, these two facts are the strongest evidence that the universe started in a big bang. The explanation of the radiation is that if the universe was originally very hot, it would have produced lots of high-energy radiation, which, as it travelled through the universe for the next 10bn years, gradually decreased in energy (think about the expansion of hot air, which cools as it expands), eventually ending as the extra bit of static on a television set. The explanation of the helium, according to big bang theory, is that during the first three minutes of its existence, the universe effectively acted as a nuclear fusion reactor. Straightforward calculations show that helium would have been cooked from hydrogen in roughly the right amounts to match up with what we see today.
The discovery in 1965 of cosmic background radiation led to virtually universal acceptance of big bang theory. Nevertheless, for almost three decades, our knowledge of the history of the universe remained hazy. We didn't know when it began—estimates of its age ranged from ten to 20bn years—and we didn't know how it would end. Since gravity must be gradually slowing down the expansion, the two possibilities seemed to be, in Robert Frost's words, fire or ice. Either gravity would eventually stop the expansion, the universe would start to contract and eventually end in a reverse big bang, or "big crunch"; or the universe would gradually decelerate but the expansion would never actually stop, leading to an empty stage of burned-out stars and black holes.
A further problem was that it was difficult to see how the richly varied universe we see around us could have come into existence. The uniformity of cosmic background radiation showed that shortly after the big bang the universe was filled with gas with a remarkably constant density. But a perfectly uniform sea of gas could not have led to the lumpy universe we see today, a sea of galaxies, stars, planets and daffodils. Astronomers generally assumed that there must originally have been tiny differences in the density of the gas from place to place, which—given 10bn years and the effect of gravity—would have led to the lumpy universe we see around us. However, that should have produced small variations in cosmic background radiation—and for almost three decades nobody could find any evidence for these.
Yet astronomy remained fertile during this period, even if cosmologists could not answer many of the big questions. Apart from the attempts to detect variations in the cosmic background radiation, there were many attempts to determine the ultimate fate of the universe. In principle, answering this question should have been quite simple, because the ability of gravity to stop the expansion depends on only one number: the average density of the universe. Unfortunately, the uncertain contribution of dark matter—matter which does not emit any light and which we can detect only through its gravitational effect—made this number very difficult to measure. Cosmologists also tried another simple method, called the "standard candle" technique. For a technical reason, the geometry of the universe and its ultimate destiny are connected, the answer to one providing the answer to the other. The standard candle technique, which was first used by Edwin Powell Hubble in the 1930s, is a way of measuring the geometry of the universe. Suppose one finds a class of object such that all objects in it have the same luminosity—call these standard candles. By observing how faint these objects are at different distances, one can, in principle, determine the geometry of the world in which the candles are found, from the Euclidean geometry of our everyday world which makes real candles obey an inverse square law (a candle twice as far away is four times as faint, three times as far away nine times as faint and so on) to the broader geometrical possibilities of an expanding universe. Hubble selected the brightest galaxies in clusters as his standard candles, and these were the standard candles of choice for five decades. However, in the 1970s it became clear that galaxies far enough away for geometric effects to be evident were also being seen so far back in time that they were no longer standard candles because of the evolution in their populations of stars; evolution trumped geometry and the technique fell out of favour.
The revolution in cosmology
The first shot in the cosmological revolution was fired in 1998 when an international team published the results of a new attempt to measure the geometry of the universe using the standard candle technique, although this time with supernovae rather than galaxies as the standard candles. A supernova—the aftermath of an exploded star—is not the most obvious standard candle, because after the explosion it increases in brightness by a factor of 1bn before gradually fading away over a few months. But the supernovae in one particular class do appear to all have the same luminosity when they reach the point of maximum brightness—and so if one can catch a supernova at this point, and separate the light of the supernova from the light of the surrounding galaxy, both of which were difficult but possible with the new instruments available in the 1990s, one has a standard candle.
The team started with the usual aim of using the standard candles to determine the geometry of the universe, but when they plotted supernova brightness against distance they found that the most distant supernovae were fainter than predicted by the standard models. They found that they could get agreement with their results only if the universe's expansion were actually accelerating, which had the revolutionary implication that there must be a force acting on the universe that is counteracting the decelerating effect of gravity.
I heard about this discovery the way I seem to hear about most cosmological discoveries—listening to the Today programme while doing the washing-up. My first thought was that there was something wrong with the observations. The big problem with cosmological experiments is that one is always looking for signals that are only just detectable over the experimental noise, which makes it difficult to be sure there are no other effects causing the signal.
The thing that began to convince me, and I suspect most cosmologists, that this result might be true was a second discovery two years later. The expected variations in the brightness of the cosmic background radiation had been detected in 1992 by a Nasa satellite, finally solving the embarrassing problem of the origin of the lumpy universe. The most recent image of the cosmic background radiation, made by a new Nasa satellite, the Wilkinson Microwave Anisotropy Probe (WMAP) and published in 2003, is shown above. The variations in brightness of the radiation have been exaggerated in the image; the true variation in brightness from place to place is only about one part in 100,000, showing that shortly after the big bang, the universe was a remarkably homogeneous place. This is an image of the deep past, a direct picture of what the universe looked like only 400,000 years after the big bang—a mind-blowing concept which may excuse some of the silly statements that were made by scientists when the original image was released to the media ("it is like looking at the face of God" and so on). The original 1992 image, one of the few astronomical images ever to make the evening news, was only a broad-brush image of this radiation, and during the 1990s many teams tried to obtain a much more detailed picture. When one of the teams succeeded in 2000, they discovered that this picture contained the answer to another important cosmological question.
This discovery was not unexpected, because in the 1960s theorists had suggested that the minuscule density variations in the early universe were produced by sound waves in the gas, and that these sound waves would have certain natural wavelengths imposed by the size of the universe (think of the natural harmonies of any musical instrument, which are set by the size and shape of the instrument). The most important of these sound waves—the dominant note of the universe—had a wavelength of about 600,000 light years.
Although this discovery was not unexpected, the team were able to use their results to carry out a remarkable piece of high-school trigonometry. Since they knew the physical scale of these sound waves, the angle they make on the sky and the distance the cosmic background radiation has travelled, they were able to construct a huge triangle spanning the observable universe. Because the sum of the angles in a triangle depends on the curvature of space, they were then able to estimate the universe's geometry—which turned out to be indistinguishable from the flat Euclidean geometry of our everyday world. Even more remarkably, by combining their geometrical measurement with the best measurements of the average density of the universe, they discovered that their results too implied that the expansion of the universe is accelerating.
All the cosmological experiments since the turn of the millennium have confirmed these two results. The difference between the state of cosmology in the 1990s and today is not that there are more answers, but that now the answers all agree. Here they are.
On the largest scale, the universe has a flat Euclidean geometry. It is 13.7bn years old, give or take about 0.4bn. About 16 per cent of the matter in the universe is the stuff that makes up our everyday world—protons and neutrons—with the rest being an unknown kind of dark matter. There is a mysterious force in the universe counteracting the effect of gravity and causing the universe's expansion to accelerate. The other galaxies will therefore continue to move away from our own; the sun will die but new stars will continue to form out of gas; after all the gas is used up, the final generation of stars will gradually flicker out, ending their lives as black dwarves, neutron stars or black holes—and so the answer to Robert Frost's question is that the fate of the universe is ice.
Thus we now have much more precise knowledge of the history and basic properties of the universe, but we are also faced with two huge new questions. We do not know the composition of the dark matter, although particle physicists have some speculative ideas. We have even less understanding of the mysterious large-scale force. This is sometimes called the "cosmological constant" or sometimes "dark energy." Quantum field theory, the theory at the basis of particle physics, implies that there should be a force causing the universe to accelerate, but its value should be 10120 (1 followed by 120 zeroes) times greater than what we actually observe. The existence of a force at the level we observe is a problem to which neither particle physicists nor astronomers have any plausible answers.
The "big science" problem
So what next? The cosmic background radiation allows us to look back to a time only 400,000 years after the big bang. We cannot directly observe events before this because the universe was ionised, which obscures our view in the same way that the centre of the sun, a ball of ionised gas, is hidden. Nevertheless, in the same way that we can infer what is going on in the centre of the sun by studying its surface, we can infer what happened earlier in the history of the universe by observing the details of the cosmic background radiation.
In the long term, there is the possibility of overcoming the ionised gas problem by using telescopes to detect gravitational waves, which are not scattered by ionised gas. The first gravitational-wave telescopes are now in operation, and so there is now a prospect of looking much further back, possibly to the time at which the sound waves were generated—the so-called "period of inflation."
The best bet for discovering more about dark energy seems to be more normal telescopes. For example, the US-led Dark Energy Survey will use a telescope of the same size as the one I was using back in the 1990s. The DES team will observe approximately 300m galaxies and use a variety of techniques to measure how the acceleration caused by dark energy has changed during the history of the universe. Although this is unlikely to solve the ultimate mystery of dark energy, it will still be a valuable step forwards.
The worrying thing about this list of projects is that they all fall into the category of Big Science. In a recent influential article, the theoretical cosmologist Simon White highlighted some of their dangers. Some of these are fairly obvious sociological problems that are a routine fact of life in other scientific disciplines: disruptive intergenerational squabbling and so on. However, these problems are probably inevitable, and we will just have to try to mitigate them. The most useful of White's warnings is against big cosmological projects with only a single goal.
Suppose all the world's nations get together to build a telescope with the aim of measuring a single cosmological parameter—call it the Babel Telescope. Billions of euros go into this project, but it doesn't quite work. Imagine the effect on the public. Whereas the taxpayer has always been fairly forgiving of satellites that explode on launch or missing Mars missions, I suspect a big failure like this would lead to a shattering loss of credibility for cosmology as a whole. This is not too far from the situation now in particle physics with the Large Hadron Collider, due to begin smashing protons into one another later this year, and arguably at least one huge astronomy project on the drawing board: the "European Extremely Large Telescope," which really has only one goal: looking for earth-like planets around other stars.
White's prescription is that any big project should always have multiple aims, so if the big cosmological goal does not work out, there are still lots of small-scale astronomy projects left to do. For an astronomer like me, whose heart is still with high-redshift radio galaxies, this is balm for the soul.
