Dark matter and dark energy

Here is one of the strangest things science has found. Almost all of the universe is missing.

Stars, planets, gas, dust, people: everything made of atoms adds up to about a twentieth of what is out there.

About a quarter is something called dark matter. It gives out no light, and we cannot see or touch it. But it has gravity. Galaxies spin so fast that they should fly apart, and something unseen is holding them together.

The rest, more than two thirds, is called dark energy. It does the opposite of gravity. It is pushing the universe apart faster and faster.

The names are labels, not explanations. “Dark” here means “unknown”. Nobody yet knows what either of them is. Working that out is one of the biggest unsolved problems in science.

The two have little in common except the word “dark” and the fact that neither is understood. The evidence for each is strong. What they are is unknown.

	Share of the universe today	What it does	How we know
Ordinary matter	About 5 per cent	Makes stars, planets, gas and us	We see it
Dark matter	About 27 per cent	Pulls, like any matter, but gives out and blocks no light	Its gravity
Dark energy	About 68 per cent	Makes the expansion of the universe speed up	The expansion history

Dark matter

Galaxies turn too fast. In the Solar System, planets far from the Sun move slowly, because nearly all the mass is at the centre. A galaxy’s light is also concentrated towards its centre, so stars at the edge should move more slowly than stars further in. In the 1970s Vera Rubin and Kent Ford measured them, and they do not. The speeds stay level right out to the edge and beyond. For that to happen, each galaxy must sit inside a much larger ball of unseen mass.

Clusters are too heavy. In 1933 Fritz Zwicky found that the galaxies of the Coma cluster move so fast that the cluster should have scattered long ago, unless it holds far more mass than its galaxies can account for. The hot gas between the galaxies, found later, is not nearly enough.

Light bends too much. Mass bends light. A cluster of galaxies stretches the images of galaxies behind it into arcs, and the amount of bending gives the cluster’s total mass with no need to see it. The answer is about five or six times the mass of everything visible.

One collision. The Bullet Cluster, about 3.8 billion light years away, is two clusters that have passed through each other. Their hot gas, which is most of their ordinary matter, collided and was left behind in the middle. Lensing shows that most of the mass went straight on with the galaxies. The mass and the ordinary matter are in different places.

The oldest light. The pattern of patches in the cosmic microwave background depends on how much matter there was that light could push about and how much it could not. It gives the same answer: about five parts dark to one part ordinary.

What it might be. Not dim stars, planets or black holes of ordinary origin: searches have ruled those out as the main part. The leading idea is a kind of particle not yet discovered, one that hardly interacts with atoms or light at all. Detectors deep underground have waited for decades for one to strike an atom. So far none has been seen, and the simplest candidates are being ruled out one by one.

A different idea is that there is no extra matter and that gravity behaves differently over great distances. Changed laws of gravity can match the spin of single galaxies well. They have not managed to explain the Bullet Cluster and the oldest light at the same time, so most astronomers favour dark matter. The question stays open until something is detected.

Dark energy

The discovery. The gravity of everything in the universe should slow the expansion down. In 1998 two teams set out to measure by how much. They used one kind of exploding star, a white dwarf blown apart, which always reaches nearly the same true brightness and so shows its distance. The far ones were fainter than they should have been. The expansion is not slowing. It has been speeding up for about the last 6 billion years. The work won a Nobel Prize in 2011.

What it might be. The simplest answer is that empty space itself has a fixed amount of energy in every cubic metre. As space grows there is more of it, so its push comes to dominate. Einstein’s equations have room for exactly such a term. The trouble is that attempts to work out its size from the physics of particles give answers that are wrong by an absurd margin.

Is it constant? In 2024 and 2025 a survey called DESI, which is mapping tens of millions of galaxies, reported signs that dark energy may have weakened over time. The evidence is not yet strong enough to count as a discovery. The Euclid space telescope, launched in 2023, and other surveys are measuring the same thing independently.

Why it matters. If dark energy stays constant, the expansion goes on speeding up for ever, and distant galaxies will one by one pass out of sight. If it changes, the far future is not settled.

Two unknowns, measured well. The amounts of dark matter and dark energy are known to a per cent or two. What either of them is has not been settled by any experiment. This level sets out the numbers, the searches that have come back empty, and the one measurement that may be showing something new.

The amounts. The Planck satellite’s reading of the oldest light gives the density of each ingredient as a fraction $\Omega$ of the total, multiplied by $h^2$, where $h$ is the Hubble constant divided by 100 km/s per megaparsec.

Quantity	Planck 2018 value
Dark matter, $\Omega_c h^2$	0.120 ± 0.001
Ordinary matter, $\Omega_b h^2$	0.0224 ± 0.0001
Hubble constant, $H_0$	67.4 ± 0.5 km/s per megaparsec
All matter, $\Omega_m$	0.315 ± 0.007

Three things follow, each worked out here from those figures. Dividing the first row by the second gives 5.4 parts of dark matter to one of ordinary matter. With $h = 0.674$, dark matter is 26.4 per cent of the total and ordinary matter 4.9 per cent. In a flat universe, which is the model Planck’s data fit, whatever is not matter is dark energy: one minus the last row, or 68.5 per cent. The familiar 27 per cent for dark matter is a rounded figure from earlier data. The 2018 analysis gives about 26.

How thin it all is. The total density of a flat universe is fixed by the expansion rate:

$$\rho_c = \frac{3 H_0^2}{8 \pi G}$$

With Planck’s $H_0$ this comes, by our own sum, to $8.5 \times 10^{-27}$ kg/m³, the mass of five protons in each cubic metre. Ordinary matter is one proton in every four cubic metres. Dark energy is $5.8 \times 10^{-27}$ kg/m³, which as energy is $5.3 \times 10^{-10}$ joules in a cubic metre. It dominates only because space is so empty of everything else.

Dark matter: looking for a heavy particle. If dark matter is a particle, the Earth is moving through a thin gas of it, and now and then one should strike the nucleus of an atom. The most sensitive search so far is LZ, which watches seven tonnes of liquid xenon 4,850 feet underground in South Dakota, where the rock keeps out cosmic rays.

Experiment	Exposure	Result	Tightest limit on the cross-section
XENONnT, 2023	About 1.1 tonne-years	No significant excess	$2.58 \times 10^{-47}$ cm² at a particle mass of 28 GeV/c²
LZ, 2024	4.2 tonne-years, from 280 days of running	No excess over the expected background	$2.2 \times 10^{-48}$ cm² at 40 GeV/c²

The cross-section is the effective target area a nucleon presents to the particle. The smaller the limit, the more feebly the particle must interact if it exists. A mass of 40 GeV/c² is about 43 times the mass of a proton, by our own sum. These limits apply to particles of 9 GeV/c² and heavier, the kind long known as WIMPs, for weakly interacting massive particles. They say nothing about much lighter ones.

Dark matter: looking for a very light particle. The axion is a particle proposed for a different reason, to solve a puzzle about the strong nuclear force, and it would also serve as dark matter. In a strong magnetic field an axion can turn into a photon of frequency

$$f = \frac{E}{h}$$

where $E$ is the axion’s energy, nearly all of it the energy of its mass, and $h$ here is Planck’s constant. The ADMX experiment is a 136-litre copper-plated cavity inside a 7.6 tesla magnet, tuned like a radio to pick up that photon. The power expected is of the order of $10^{-23}$ watts.

In 2020 ADMX reported a search for axion masses from 2.81 to 3.31 millionths of an electronvolt. By our own sum that is radio frequencies from 679 to 800 MHz. It found nothing, and ruled out the benchmark axion models across that window. The window is narrow. The same paper notes that one line of argument puts the axion’s mass at about a millionth of an electronvolt or more, so a great deal of the range is still to be searched.

The two kinds of particle would be very different in number. The ADMX analysis takes the dark matter round us to have a density of 0.45 GeV in each cubic centimetre. Worked out here, that is about 11 particles a litre if each has a mass of 40 GeV/c², or about 150 million million in every cubic centimetre if each is an axion of 3 millionths of an electronvolt.

Where that leaves dark matter. Every piece of evidence for dark matter is still its gravity. No particle has been detected in a laboratory. The searches have not shown that there is no particle. They have shown that the simplest heavy candidates interact more feebly than was once hoped, and they have opened a first narrow window on the light ones.

Dark energy: the equation of state. How anything affects the expansion depends on one number, the ratio of its pressure $p$ to its energy density:

$$w = \frac{p}{\rho c^2}$$

As the universe grows by a scale factor $a$, a component with constant $w$ thins out as

$$\rho \propto a^{-3(1+w)}$$

and the expansion speeds up or slows down according to

$$\frac{\ddot a}{a} = -\frac{4 \pi G}{3}\left(\rho + \frac{3p}{c^2}\right)$$

Matter has $w = 0$, thins as $a^{-3}$, and slows the expansion. Anything with $w$ below $-\tfrac{1}{3}$ makes the bracket negative and speeds it up. A cosmological constant, energy belonging to space itself, has $w = -1$ exactly, so its density never changes.

Two dates follow if dark energy is a constant, both worked out here with Planck’s figures. Matter thins and dark energy does not, so the two were equal when the universe was smaller by a factor of 1.30, about 3.5 billion years ago. The expansion began to speed up earlier, when matter had fallen to twice the density of dark energy: a factor of 1.63, about 6.1 billion years ago.

Is it a constant? The DESI survey measures how the distances to galaxies have grown, using a pattern in the way galaxies cluster as a standard ruler.

Result	Data	Finding
DESI 2024	Over 6 million galaxies and quasars	With $w$ held constant, DESI alone gives $w = -0.99$, +0.15 −0.13. With $w$ allowed to change, a changing value is preferred at 2.6 sigma with the oldest light added, and at 2.5, 3.5 or 3.9 sigma with one of three supernova catalogues added as well
DESI 2025	More than 14 million	The same preference at 3.1 sigma with the oldest light, and from 2.8 to 4.2 sigma depending on the supernova catalogue

In both, the preferred values have $w$ above $-1$ today. By the thinning rule above, that would mean the density of dark energy is now slowly falling.

This is not yet a discovery. Physicists usually ask for 5 sigma, and here the figure depends on which supernova catalogue is used. DESI’s own data by themselves are well described by a constant. The DESI team’s wording is that, unless one of the data sets has an error nobody has found, the standard model is being challenged and a changing dark energy is a possible answer.

The size problem. Even a perfectly constant dark energy would not be understood. Quantum physics says empty space should have energy, and the natural estimate is enormous. Counting the contributions up to the highest energy at which present theory might be trusted gives about $2 \times 10^{109}$ joules in a cubic metre. The measured figure above is $5.3 \times 10^{-10}$. The two differ by about 120 powers of ten. The estimate could be cut off at a lower energy, but even well-tested physics gives contributions tens of powers of ten too large. No known principle makes them cancel.

Where it stands.

• The amounts of dark matter and dark energy are measured to a per cent or two.
• Dark matter has been detected only through gravity. The laboratory searches so far have found nothing.
• Dark energy fits a constant with $w = -1$ to within the errors of any single experiment. Combined data sets hint at change, at a level short of a discovery.
• Nobody can yet explain why the energy of empty space is as small as it is.