It turns out that there is another, arguably simpler explanation that is well supported by many other observations. It is that the entire universe itself is expanding! As I will explain below, this expansion means not only that we should see every other galaxy moving away from us, but that observers in another galaxy should see exactly the same thing. In a uniform expanding universe, every observer sees herself at the center of the expansion, with everything else moving outwards from her.
This statement forms the basis of our current theories of the structure and history of the universe. The study of the overall structure of the universe is called cosmology. The theory that has come to dominate cosmology since Hubble's observations goes by several names, but is most commonly known as the big bang model. (As I will explain later, this name is somewhat misleading, but owing to its widespread acceptance I will continue to use it.)
This paper describes the big bang model. Section II describes what it means to say the universe is expanding, and subsequent sections address some questions that commonly arise in connection with the model.
In Section III I discuss whether the universe is infinite or finite. While we don't yet know the answer to this question, Einstein's general theory of relativity predicts that finite universes contain a larger density of matter than infinite ones, so by measuring the density in the universe we could in principle be able to make the determination. I'll explain why this method hasn't yet worked. I conclude this section by describing what it would mean for the universe to be infinite or finite.
In Section IV I talk about the origin and history of the universe. As the universe expands and galaxies move apart from each other the average density is decreasing. If we extrapolate the expansion backwards we conclude that there was a time roughly 14 billion years ago when the density was nearly infinite. In this section I briefly outline the history of the universe from that time to the present.
In Section V I continue the story, describing what relativity theory predicts will happen to the universe in the future. The two possibilities are that the universe will continue to expand forever or that it will eventually slow down and begin contracting. Which of those will happen depends on the density of energy in the universe and on what type of energy it is. I describe those conditions and conclude by describing these two scenarios.
The paper is followed by a series of endnotes that discuss other issues, including evidence for the big bang model as well as possible problems with it and the proposed solutions. It is not necessary to read the endnotes to understand the rest of the paper.
Think of the universe as a rubber sheet being stretched out. (If you are comfortable with visualization in three dimensions you can imagine a raisin cake expanding instead, but for the purpose of illustration I will stick with the two dimensional case.) Now imagine that there are thumbtacks stuck into the rubber at various points representing galaxies. (In the raisin cake analogy these would be the raisins.) As the rubber (the universe) is stretched (expands), the thumbtacks (galaxies) all get farther apart. Note that I haven't said anything yet about how big the rubber sheet is. For all we know it might be infinite. (This point will be addressed in a later section.) What I mean when I talk about expansion is that the rubber is being stretched out, causing the distances between the thumbtacks to increase.
To see what this expansion should look like to us, imagine an observer sitting on one of the thumbtacks. This observer imagines himself to be at rest and measures all movement relative to his thumbtack (galaxy). Since the distance between any two thumbtacks is increasing, it will appear to him that all the other ones are moving away from him. How fast will another thumbtack appear to move? That depends in part on how fast the rubber sheet is being stretched out, i.e., how fast the universe is expanding. In addition, however, the apparent speed of the other thumbtacks is also dependent on their positions relative to the observer. The nearby thumbtacks will appear to be moving away very slowly, whereas the distant ones will appear to be moving away much faster. To see why this is so, suppose the rubber sheet doubles in size in one second.
The thumbtack that began one foot away from you is two feet away, meaning it appears to have moved by a foot. Its apparent velocity is therefore 1 foot per second. In the same time the thumbtack that started out three feet away also ends up twice as far away (six feet), but this means that it appears to have moved away at three times the speed of the first thumbtack (three feet per second). In terms of the expanding universe, this means that not only will every galaxy appear to be moving away from us, but the speed with which it does so will be directly proportional to its distance from us. A galaxy that is four million light years away will have twice the apparent velocity of one that is two million light years away.
This pattern is precisely what Hubble observed. Not only did he see that all distant galaxies are moving away from us and that the more distant ones are moving away more rapidly, but he found that the rate at which they were receding from us was proportional to their distance from us. In short, his observations exactly matched what we just predicted for an expanding universe. This proportionality is known as Hubble's Law.1
A problem arises when we consider an expanding universe. Suppose everything in the universe were to double in size. The distances between galaxies would double, the size of the Earth would double, the size of all our meter sticks would double, and so on. It would seem to an observer (who will also have doubled in size) as if nothing had happened at all. So what do we mean by saying the universe expands?
In fact, not everything grows as the universe expands. In the example of the rubber sheet, the distance between thumbtacks keeps increasing but the thumbtacks themselves remain the same size. Similarly, while distant galaxies are pulled away from each other by the expansion, smaller objects like meter sticks, people, and the galaxies themselves are held together by forces that prevent them from expanding. So we expect that billions of years from now galaxies will still be roughly the same size they are today, but the distances between them will on average be larger.
We believe that the universe is governed by Einstein's theory of general relativity, which among other things addresses such matters as the overall structure of the universe. In the early 1920s Alexander Friedmann showed that using two assumptions (which I discuss below), the equations of general relativity can be solved to show that a finite universe must have a larger density of matter and energy inside it than an infinite universe would have.2  There is a certain critical density that determines the overall structure of the universe. If the density of the universe is lower than this value, the universe must be infinite, whereas a greater density would indicate a finite universe. These two cases are referred to as an open and closed universe respectively.3
The critical density is about 10-29 g/cm3, which is equivalent to about five hydrogen atoms per cubic meter.4  This may not seem like a lot; by comparison the density of water is roughly 1 g/cm3 or about 500 billion billion billion hydrogen atoms per cubic meter. However, we live in a very dense part of the universe. Most of the universe is made up of intergalactic space, for which a density as low as the critical density is plausible.
Aside from the theory of relativity itself, Friedmann's other assumptions in deriving his results were that the universe was the same everywhere ("homogeneity") and looks the same in all directions ("isotropy"). Of course in reality the universe is not the same everywhere. I already mentioned that the Earth is much more dense than space. However, if I measure the average density in our galaxy it will be about the same as the average density in other galaxies like it, and the number of galaxies per unit volume should be roughly the same in different parts of the universe. When you average over large enough regions these assumptions seem to match our observations. Individual galaxies differ from one another in some of their specific properties, but on average their properties don't appear to change from one region of the sky to another. Nonetheless, the ideas of homogeneity and isotropy are still assumptions. We can probably only see a tiny fraction of the universe and we have no guarantee that the parts we cannot see look like the parts we can. The big bang model assumes that these properties hold throughout the entire universe, and we will continue to use that assumption throughout the rest of this paper. (My follow-up paper, Beyond the Big Bang: Inflation and the Very Early Universe, includes a brief discussion of the possibility that the universe is not homogeneous on scales bigger than we can see.)
So we should be able to answer the question of the universe being infinite or finite by measuring the density of everything around us and seeing whether it is above or below the critical value. This is true in principle, and measuring the average density of the universe is an active field of research. The problem is that the measured density turns out to be very close to the critical density. The theory of inflation, our best theory of what happened in the fraction of a second after the big bang, predicts that the actual density of the universe is likely to be so close to critical density that we may never be able to measure whether it is above or below. Another way of saying that is to say that if the universe is finite, it is probably so much bigger than the part of it we can observe that it may always look infinite to us.
To recap, one of the assumptions of the standard big bang model is that the universe is more or less homogeneousthe same everywhere. As far as we can see, which is billions of light years in every direction, this assumption appears to be correct. Under this assumption general relativity says that whether the universe is infinite or finite depends on its density. Measurements of the density of the universe show it to be so close to the critical density that we can't tell if it is above or below.
Given our uncertainty about this question, I will say a few things about what it would mean if the universe is infinite or finite and how those two possibilities relate to the idea of the universe expanding.
An infinite universe is in some ways easier to imagine than a finite one. Since the universe is supposed to be everything that exists, it seems intuitive that it should go on forever. Of course an infinite universe is impossible to picture, but we can get at what it means by saying that no matter how far you go there will always be more space and galaxies. It is hard, however, to reconcile this picture with the idea that the universe is expanding. If it's already infinite, how can it expand?
To see how, remember that by expansion we mean that the distance between galaxies is increasing. Suppose right now there is a galaxy every million light years or so. After a long enough time this infinite grid of galaxies will stretch out so that there is a galaxy every two million light years. The total size of the universe hasn't changedit's still infinitebut the volume of space containing any particular group of galaxies has grown because the separation between the galaxies is now larger.5
What about a finite universe? This phrase sounds like a contradiction because if the universe ends somewhere then we would naturally want to know what was beyond it, and since the universe includes everything, whatever is beyond that edge should still be called part of the universe. The resolution of this paradox is that even if the universe is finite, it still doesn't have an edge. If I head off in one direction and resolve to keep going until I find the end of the universe, I eventually find myself right back where I started. A finite universe is periodic, meaning that if you go far enough in any direction you come back to where you started.
Trying to picture a closed (finite) universe is in some ways even harder than trying to picture an open (infinite) universe because it is easy to mislead yourself. For example, people often compare a two-dimensional closed universe to the surface of a balloon. This analogy is helpful because such a surface has the property of being periodic in all directions, and it is easy to picture the expansion of such a universe by imagining the balloon being blown up. In fact, this analogy is like the rubber sheet analogy I used before, except now the sheet has been wrapped up to form a sphere. The problem is that this picture immediately leads to the question of what is inside the balloon.
This question comes from taking the analogy too literally. Nothing in general relativity says that a two-dimensional closed universe would have to exist as a sphere inside a three-dimensional space; the theory only says that such a universe would have certain properties (e.g. periodicity) in common with such a sphere. For this reason I think it is useful to keep the balloon in mind as a convenient analogy but it is ultimately best to think of the closed universe as a three-dimensional space with the strange property that things which go off to the right eventually come back again from the left.
What does expansion mean in a closed universe? Since this universe has a finite size, it makes sense to talk about that size increasing. Again suppose that there is now a galaxy every million light years. Suppose also that if I were to head off in a straight line I would travel 100 billion light years before coming back to where I started, passing about 100,000 galaxies on the way. If I take the same journey billions of years later, the number of galaxies won't have changed but the distances between them will have doubled, so the total distance for the round trip will now be 200 billion light years.6
Given that the universe is growing, the question of whether the expansion started at some point in the past inevitably arises. Our current theories say the expansion did have a beginning. This section discusses why we believe this and what it means to even say so. It also contains a brief outline of the history of the universe from that beginning to the present day.
Having defined the moment of the big bang in this waythe time when all distances between objects were zeroI am not going to talk about that time. A point of infinite density, known in physics as a "singularity," makes no sense. Moreover, our current theories do not predict that such a moment occurred in the past. Our best physical theories, including general relativity and quantum mechanics, stop working when we try to describe matter that is almost infinitely dense. That word "almost" is important. The theories don't simply break down at the instant of the big bang singularity; rather, they break down a short time afterwards when the density has a certain value called the Planck density.
The Planck density, which is the highest density we can hope to describe with our current physics, is over 1093 g/cm3, which corresponds to roughly 100 billion galaxies squeezed into a space the size of an atomic nucleus. For virtually any application we can imagine this limitation of our theories is completely irrelevant, but it means we can't describe the universe immediately after the big bang. We can only say that our current model of the universe begins when the density was somewhere below the Planck density and we can say virtually nothing about what the universe was like before that. We therefore take as our initial condition a universe at or just below the Planck density, and any questions about the instant of the big bang itself are eliminated from consideration.
Is this a cop-out? It certainly is. Physicists have not given up on understanding what happened before this time, but we admit that right now we have no theory to describe it. Many people are working to develop such a theory, but until that happens we are left having to start our description of the universe when the density was large but still finite.
Once we impose this limitation on ourselves, our picture of the universe works equally well for an infinite or a finite universe. If the universe is finite then it may very well have been extremely small at the moment when the density was at the Planck level. If the universe is infinite then it was also infinite at that early time. The density was enormous and the distances between particles vanishingly small, but that dense mass of particles went on forever.
At the moment when the density of matter equaled the Planck density, the universe consisted of a hot soup of elementary particles. When I say this medium was hot that means that the particles, on average, had very high energies. We don't know what types of particles existed at that moment because we can't reproduce such high densities and temperatures in a lab. As the universe expanded, the density and temperature of this mixture decreased and within about a second the universe would have thinned out and cooled to roughly the highest density and temperature we can create artificially. By that time all of the fundamental particles familiar to physicists, such as quarks, electrons, and photons, were present. Today these particles are mostly combined into larger units such as atoms, molecules, penguins, and so on, but at the extremely high temperatures of the early universe they remained separate. If several particles were to have combined into a more complicated structure such as an atom they would have been instantly ripped apart in collisions with the high energy particles flying around everywhere.
After about one second quarks combined into protons and neutrons. A few minutes later the protons and neutrons combined into atomic nuclei in a process referred to as nucleosynthesis. Hundreds of thousands of years later these protons and neutrons combined with electrons to form atoms. This last process is called recombination (despite the fact that particles had presumably never been bound into atoms before).
In the period of recombination the universe was still almost perfectly homogeneous, meaning that the density was the same everywhere. While the density still is the same everywhere when averaged over huge regions of space, it certainly varies locally. The density of the Earth is vastly larger than the density of interstellar space, which is in turn much greater than the density of intergalactic space. In contrast, the difference in density between the most and least dense regions at the time of recombination was about one part in 100,000. Between then and now the clumping of matter into galaxies, stars, etc. took place.
The mechanism by which this clumping occurred is fairly simple, although its details continue to be studied and debated. At the time of recombination the universe consisted of a nearly uniform hot gas with regions very slightly denser than the average and others very slightly less dense. If the density had been exactly the same everywhere then it would have always stayed that way. However, a region slightly denser than the surrounding gas would have a stronger gravitational attraction, and mass would tend to flow into it. This process would make this region even denser, causing it to attract matter even more strongly. In this way the almost uniformly dense universe gradually became less and less uniform, resulting in the dense clumps of matter we see around us now. On a fairly large scale these clumps make up galaxies, and matter that clumped on a smaller scale makes up the stars inside those galaxies. A very small portion formed into smaller objects orbiting around those stars and a small portion of that matter formed into people reading physics papers on the Internet.
Our view of the future of the universe has changed in the last few decades. The next paragraph is what I wrote when I originally put this paper out in 2000. I've kept it so you can see how much has changed since then. After that I give a more up-to-date description.
We know from general relativity that expansion of the universe is slowed down by the mutual gravity of all the matter inside it. Whether or not the expansion will continue forever depends on whether or not there is enough matter in the universe to reverse it. If the density of matter in the universe is less than a certain critical value, then the universe will never stop expanding. If, on the other hand, the density of matter is greater than the critical value, then the pull of gravity will eventually be strong enough to stop the expansion and the universe will begin contracting. In Section III we saw that whether or not the universe is finite or infinite depends on whether the density of matter is above or below a critical value. That value turns out to be exactly the same as the critical value that determines whether or not the expansion will reverse. In other words, general relativity says that an open (infinite) universe will expand forever and a closed (finite) universe will eventually recollapse.8In 1998 and 1999 two groups independently published measurements of how the expansion rate of the universe is changing over time. By measuring how rapidly gravity is slowing the expansion, they should have been able to determine whether we are above or below the critical density. To everyone's surprise, their results showed that distant galaxies are not slowing down as they move away from us; they are speeding up. It's as if you walked outside and threw a ball up in the air, and instead of slowing to a stop and coming back down it kept moving up, faster and faster. That's not how we normally understand gravity. What this implies is that in addition to the ordinary, attractive force of gravity between galaxies, something in the universe is exerting an even stronger, repulsive gravitational force. As of 2018 we still don't know what that something is, so we just call it dark energy. (It's an unfortunate name because it's so easy to confuse with dark matter, which is a completely different thing.) These measurements have been independently confirmed in a number of ways, and it seems pretty certain now that about 2/3 of the energy in our universe is this mysterious dark energy. (When this paper was first written in 2000 dark energy was still so new and uncertain that it only merited one brief footnote in the original paragraph above.)
So what does this imply about the future of the universe? If most of the gravity in the universe is repulsive then the galaxies will never stop and come back together, regardless of their density. In other words, the idea that a closed universe recollapses and an open one expands forever is only true for a universe dominated by ordinary matter. A universe dominated by dark energy should keep expanding forever. There is an important caveat, however, which is related to the fact that we still don't know what dark energy is. If it someday decays and turns into ordinary matter then the fate of the universe will once again depend on whether it is closed (finite) or open (infinite). In sum, if the universe remains dominated by dark energy forever then it will never stop expanding. If it is open and someday becomes dominated by matter then it will also never stop expanding. But if it is closed and someday becomes dominated by matter, then it will eventually recollapse. So what would each of those scenarios look like?
If the universe expands forever, the clusters of galaxies in it will move farther and farther apart. Eventually each galaxy cluster will be alone in a vast empty space. The stars will burn out their fuel and collapse, leaving nothing but cold rocks behind. Eventually these will disintegrate as well. This whole process will take an unimaginably long time but it will occur eventually, and the universe will thereafter consist of nothing but loosely spread out elementary particles. All of the energy in the universe will then be distributed in a more or less uniform way at some extremely low temperature, and as the universe continues to expand this temperature will fall and the universe will become ever more empty and cold. This scenario is sometimes referred to as the heat death of the universe.
On the other hand, if dark energy decays and the universe has a high enough density, then the galaxies will eventually start moving back towards each other. Once they are close enough together all galaxies and stars will collapse, until at some point the universe will once again consist of nothing but densely packed, highly energetic particles. Eventually all matter will be compressed to the Planck density, the density at which our current theories fail. Lacking a theory for such densities, we cannot predict what will happen then. One possibility is that the universe will bounce backindeed, perhaps it has been in a cycle of expanding and contracting forever. Then again perhaps the universe will simply annihilate itself and cease to exist. Determining which of these possibilities would occur will require the development of a theory of physics at extremely high densities.
More than any other time in history, mankind faces a crossroads. One path leads to despair and utter hopelessness. The other, to total extinction. Let us pray we have the wisdom to choose correctly.
The steady-state models were dealt their death blow with the second great piece of observational evidence for the big bang model, namely, the discovery of the microwave radiation left over from the early universe. Prior to recombination, the universe consisted of a uniform hot mixture of particles. Such a mixture emits a recognizable spectrum of radiation that, if emitted then, should still be around today. Moreover, since that mixture filled the entire universe, that radiation should have been emitted everywhere in all directions, and should thus fill all of space. In 1964 Arno Penzias and Robert Wilson discovered microwaves coming from all directions in the sky, with exactly the spectrum predicted by the theory. (The spectrum of radiation is a description of the intensity of the radiation at different frequencies.) Almost immediately after this discovery, the steady state theories were abandoned and big bang cosmology became nearly universally accepted.10
Another prediction of the big bang model concerns the relative abundances of certain light elements. According to the model, the universe started with only elementary particles that eventually formed into atomic nuclei. A hydrogen nucleus is simply a single proton, so hydrogen was the first atomic nucleus to appear . Some of the protons eventually combined with other protons and/or neutrons to form other light elements such as deuterium, helium, and lithium. The laws governing nuclear physics are fairly well understood, so physicists have been able to work out the proportions of these different elements that should have been produced. Those proportions closely match what we observe in the universe today.
The success of the big bang model required the assumption that the universe was almost exactly homogeneous (the same everywhere) at early times. If the universe had been slightly less homogeneous initially, it would look very different now, whereas if it had been perfectly homogeneous then structures such as galaxies could never have formed. Another necessary assumption is that the expansion began simultaneously throughout a very large and possibly infinite universe.
The big bang model also requires the density of matter in the early universe to have been extremely close to the critical density. If it had been too high, the universe would have recollapsed before any structure had time to form, while if it had started out too low galaxies could not have formed. I noted in endnote I that over time the universe tends to move away from the critical density. It turns out that if the universe had initially been above or below the critical density by more than one part in 1055, life as we know it could not have arisen!
These objections, while they make the theory seem strange, can be dismissed by saying that the universe just happened to start that way. Since the big bang model says nothing about how the universe got here in the first place, we have to assume some initial conditions. We are free to assume that for whatever reason the universe started out in exactly the way it had to in order to produce galaxies, stars, and ultimately you.
There is, however, another class of problems with the big bang model that cannot be explained away so easily. These problems have to do with exotic objects that should have been formed when the universe was extremely hot and dense. Our current theories predict that many different kinds of particles would have been created at those temperatures that could not be created today. Some of them would have decayed by now into normal matter and thus we would not expect to see them now, but otherscalled relic particles would be expected to be stable enough to still be present in large quantities and easily detectable. These particleswhich I won't describe in detailinclude magnetic monopoles, gravitinos, axions, and even stranger beasts such as hedgehogs, cosmic strings, and domain walls. (The last two aren't particles but large objects, but the basic idea is the same.) The fact that we don't see any of them now cannot be explained by the standard big bang model. Moreover, some of these particles, if they had been around at the time of nucleosynthesis, would spoil our successful predictions of the relative abundances of light elements (see endnote II).
Physicists have tried for decades to formulate theories that could eliminate both the questionable assumptions and the problematic particles associated with the standard big bang model. Currently the only plausible candidate is a theory called inflationary cosmology, which is widely accepted by most cosmologists to be a necessary modification of the big bang model. This theory says that there was a period of very rapid expansion in the first fraction of a second after the big bang, or more precisely, after the density fell below the Planck level. The theory of inflation is described in my paper Beyond the Big Bang: Inflation and the Very Early Universe. Here I will simply note that this rapid expansion period would have caused the universe to become almost perfectly homogeneous and almost exactly at the critical density regardless of how it started out. It would also get rid of all unwanted relic particles while still allowing for the creation of the ordinary particles that make up the universe today.
Finally I should mention the last great failing of the big bang model. Even when supplemented by inflation, big bang cosmology cannot explain why the universe is here in the first place. Inflation greatly reduces the number of assumptions you have to make about the origin of the universe. In fact some versions of inflationary cosmology suggest that the universe had no beginning but has existed forever. But whether the universe has existed forever or for only 14 billion years, the question of why it exists at all remains a mystery. Even if we could eventually come up with a set of laws that explained how the universe came into being, as some people are currently trying to do, the mystery of why those laws should exist would remain. That mystery will perhaps remain forever beyond the ability of science to explain.
1. If you know something about the theory of relativity it may occur to you that Hubble's law seems to predict that very distant objects will recede from us faster than light, whereas Einstein's special theory of relativity predicts that nothing can move faster than light. For readers who are familiar with special relativity I can note that an observer in an expanding universe is not in an inertial reference frame, and therefore the laws of special relativity do not apply. They will still be good approximations for measurements of nearby objects, but not for very distant ones. For readers not familiar with special relativity I will simply note that Hubble's law is correct and that the explanation of why this is possible requires more relativity theory than I can explain in this footnote.
2. Actually saying "matter and energy" is redundant, because according to relativity theory matter is just another form of energy, with the amount of energy corresponding to a given mass being given by the famous equation E=mc2. So from now on when I say "density of matter" I will be including all other forms of energy, such as electromagnetic radiation.
3. If the density has exactly this critical value then the universe is also infinite, but in this case it is called "flat" rather than "open."
4. Actually the value of the critical density changes with time. For a discussion of this issue see Endnote I
5. This picture of a uniform grid of galaxies is only a rough description. For example, many galaxies clump together in large groups called clusters. These clusters are held together by the mutual gravitational attraction of the galaxies so they don't grow as the universe expands. In such cases it is the distance between clusters of galaxies that grows in the way I've described.
6. The rather fanciful journey I'm suggesting is unrealistic in several ways. First of all I'm assuming that I could travel so quickly that the universe wouldn't grow much while I was making the trip. In fact even a light beam can't travel that fast and nothing can travel faster than a light beam. I also assumed for the purpose of illustration that galaxies wouldn't be created or destroyed in such a long time.
7. I'm being unrealistic when I talk about the distances between galaxies at these early times. Galaxies did not form until many millions of years after the big bang. The very early universe consisted of a dense mass of particles and the expansion of the universe at this time consisted of the distances between these particles increasing.
8. These conclusions about the future of the universe depend on an assumption that the universe is made up of ordinary matter. Recent observations suggest that the universe may instead be largely made up of a poorly understood form of matter that repels rather than attractsa kind of antigravity. If these observations are confirmed and the universe does contain such matter, then the expansion will continue forever regardless of whether the universe is infinite or finite.
9. Actually this isn't true for nearby galaxies. Having nothing to do with the expansion of the universe, galaxies have their own velocities relative to each other, known as peculiar velocities. For nearby galaxies these peculiar velocities dominate and the galaxies may be moving towards or away from us. For distant galaxies, however, the recession rate due to the expansion of the universe is so great that the peculiar velocity makes no noticeable difference.
10. The discovery of the microwave background radiation by Penzias and Wilson was a remarkable example of serendipity in science. They were doing an unrelated experiment and found that their detectors were picking up a background signal coming from all directions. It wasn't until they discussed this finding with a colleague that they understood the significance of the discovery.