Dark matter

• 1. Introduction

  ›

  Dark matter, added up from the Sun out to the orbit of Neptune, would weigh about as much as a large asteroid. That is a strikingly small amount. Yet across the universe as a whole, dark matter makes up roughly 85% of all mass. It does not interact with light, so no telescope has ever seen it. No laboratory has ever confirmed catching a single particle of it. Astronomers know it only by the pull of its gravity on things that can be seen. Who first suspected that something was missing from the cosmic ledger? What kinds of evidence forced the idea into the mainstream? And after decades of increasingly sensitive searches, why does the hunt keep coming back empty-handed?

  813341937

• 2. Dark Bodies

  ›

  In 1884, Lord Kelvin gave a series of lectures in Baltimore, and the book that grew from them asked how many stars might surround the Sun. He worked out their density from how fast nearby stars move. He then imagined a thousand million stars within one kiloparsec of the Sun. His conclusion was that a great majority of them "may be dark bodies."

  In 1906, Henri Poincaré took up Kelvin's work and used the French phrase matière obscure, meaning dark matter. He argued that the dark material must be less plentiful than the visible kind. That guess was later found to be false.

  The Dutch astronomer Jacobus Kapteyn reached for unseen matter in 1922, again using the velocities of stars. A publication by the Swedish astronomer Knut Lundmark in 1930 suggests he was the first to propose that the universe holds far more mass than anyone can observe. In 1932, the Dutch radio astronomy pioneer Jan Oort studied stellar motions near the Sun and claimed the galactic plane held extra mass. That measurement was later judged incorrect.

  The Swiss astrophysicist Fritz Zwicky made the sharper case in 1933 while working at Caltech. He applied the virial theorem to the Coma Cluster and found evidence of unseen mass, which he called dunkle Materie. He compared a mass estimate from the motions of galaxies near the cluster's edge with one based on brightness and galaxy counts. By his numbers, the cluster held about 400 times more mass than the eye could account for.

  Zwicky's figures were off by more than an order of magnitude, mostly because he used an obsolete value of the Hubble constant. His central point still held up, since most of the gravitating matter really was dark. But he pictured ordinary matter that simply gave off no light, which is not how modern theories describe it.

  In 1939, H.W. Babcock measured how the Andromeda Galaxy rotates and found its outskirts moving unexpectedly fast. He blamed light absorption inside the galaxy, or altered dynamics in the spiral's outer parts, rather than hidden matter. A year later, Oort wrote about the large non-visible halo surrounding the galaxy NGC 3115.

  2123242631

• 3. Flat Curves

  ›

  In 1974, two independent groups published papers reaching the same conclusion. One team worked in Princeton, New Jersey, where Jeremiah Ostriker and Jim Peebles were among the authors. The other worked in Tartu, Estonia, and included Jaan Einasto. Both argued that galaxies must be wrapped in halos of unseen matter.

  Their case leaned heavily on rotation curves, which plot how fast a galaxy's material orbits at different distances from the center. In the Solar System, planets farther from the Sun move more slowly, as Kepler's third law demands. A spiral galaxy's luminous mass thins out toward its edges, so the same decline in speed should appear. It does not. The curves stay flat, or even rise.

  The simplest fix is to put a great deal of non-luminous matter in the galaxy's outskirts. In optical astronomy, Vera Rubin and Kent Ford used a new spectrograph to measure the velocity curves of edge-on spirals more accurately than before.

  Radio astronomers attacked the same puzzle with a different tool, the 21 cm line of atomic hydrogen. Hydrogen gas often stretches far beyond the visible starlight, so it let them follow rotation to much greater distances. Early maps of Andromeda made with the 300 ft telescope at Green Bank and the 250 ft dish at Jodrell Bank did not show the expected decline.

  Rogstad and Shostak had used the Owens Valley interferometer in 1972 to map hydrogen in five spirals, and all five curves were very flat. In 1975, Roberts and Whitehurst traced Andromeda's rotation out to 30 kpc, far beyond the optical data. Their cumulative mass was still rising linearly at the last measurement. In 1978, Albert Bosma added more flat curves from the Westerbork Synthesis Radio Telescope.

  That same year, Steigman and colleagues published a calculation about any stable, electrically neutral, weak-scale particle left over from the early universe. It linked such a particle's mass and interaction strength to how much of it would survive today, and it became the template for the particles now called WIMPs.

  4044434546

• 4. Clusters, Lenses, Collision

  ›

  Galaxy clusters are prized by dark matter researchers because their masses can be estimated in three independent ways. The first uses the scatter in the radial velocities of member galaxies. The second uses X-rays from the hot gas between them, whose spectrum and flux reveal its temperature and density, and so its pressure. The third is gravitational lensing, which weighs a cluster without any reference to motion. The three methods agree reasonably well that dark matter outweighs visible matter by approximately 5 to 1.

  In 1987, Nick Kaiser predicted a subtle distortion in maps built from galaxy redshift surveys. Superclusters pull nearby galaxies toward them, so they look squashed along our line of sight, while voids look stretched. The effect cannot be seen in any single structure, but it shows up when many are averaged. The 2dF Galaxy Redshift Survey first measured it decisively in 2001, in agreement with the Lambda-CDM model.

  Lensing is a consequence of general relativity, in which any mass bends passing light, whatever the mass is made of. Around distant clusters such as Abell 1689, strong lensing smears background galaxies into arcs, and the geometry of the arcs gives the cluster's mass. Weak lensing causes only tiny distortions, but by examining the shear of many neighboring background galaxies, astronomers can chart the average distribution of dark matter.

  The most striking test came from the Bullet Cluster, the wreckage of a recent collision between two galaxy clusters. Lensing puts the center of mass in a different place from the center of the visible matter. In the crash, the clusters' hot gas was slowed by electromagnetic interactions, while dark matter sailed onward untouched. Theories that modify gravity generally expect lensing to track visible matter, so they struggle to explain this separation.

  7275767769

• 5. Echoes of the Early Universe

  ›

  The cosmic microwave background is very close to a perfect blackbody, yet it carries temperature ripples of a few parts in 100,000. Those ripples are a fossil record of the early universe. Back then, ordinary matter was ionized and tangled up with radiation through Thomson scattering, while dark matter ignored radiation and felt only gravity. The two kinds of matter therefore evolved differently and left different marks on the sky.

  When the ripples are broken down by angular scale, they reveal a series of acoustic peaks that are nearly evenly spaced but vary in height. The COBE satellite discovered the anisotropy in 1992, though its resolution was too coarse to resolve the peaks. The balloon-borne BOOMERanG experiment found the first peak in 2000. The WMAP satellite then mapped the pattern from 2003 to 2012, and the Planck spacecraft did so more precisely from 2013 to 2015. The detailed shape of the spectrum fits the Lambda-CDM model well and is hard for rivals such as MOND to reproduce.

  A second echo lives in the galaxies themselves. Sound-like waves in the early photon and baryon fluid left a preferred distance, so pairs of galaxies are about 1% more likely to sit 147 Mpc apart than 130 to 160 Mpc apart. Theorists predicted this feature, called baryon acoustic oscillations, in the 1990s. The Sloan Digital Sky Survey and the 2dF survey both found it in 2005.

  Then comes the matter of timing. Radiation dominated the very young universe and washed out density ripples in ordinary matter, so ordinary matter alone would not have had time to build the galaxies seen today. Dark matter is immune to radiation, so its clumps could start growing first. Their gravity formed wells that ordinary matter later fell into, speeding up the whole process. The result is a cosmic web of filaments, with superclusters at the nodes.

  986937079

• 6. What It Cannot Be

  ›

  The Lambda-CDM model divides the universe's mass and energy into 5% ordinary matter, 26.8% dark matter, and 68.2% dark energy. The 5% is the familiar stuff astronomers call baryonic, a category that includes planets, brown dwarfs, red dwarfs, white dwarfs, neutron stars, and even black holes. Dim objects of this kind are collectively called MACHOs, and some scientists once hoped they could account for all the missing mass.

  Several lines of evidence closed that door. Diffuse baryonic gas or dust in sufficient amounts would show up when backlit by stars. Big Bang nucleosynthesis, which predicts the observed abundance of the elements, allows baryons only 4-5% of the critical density, while total matter comes to about 30%. Searches for gravitational microlensing in the Milky Way excluded objects from half the Earth's mass up to 30 solar masses, covering nearly every plausible candidate. And WMAP and Planck data indicate that around five-sixths of all matter interacts mainly through gravity.

  Physicists also sort dark matter into cold, warm, and hot types. The labels describe speed rather than true temperature. They depend on the free streaming length, the distance particles drifted through random motion in the early universe before expansion slowed them. Fast particles would erase small-scale clumps, so hot dark matter predicts huge supercluster-sized pancakes forming first and fragmenting later.

  Deep-field observations show the opposite order, with galaxies first and clusters later. That points to mostly cold dark matter. It also rules out neutrinos as the bulk of it, since they travel at nearly the speed of light. A hybrid theory of mixed dark matter was popular in the mid-1990s, but it was rejected after the discovery of dark energy.

  5657595885

• 7. The Particle Hunt

  ›

  If dark matter is made of subatomic particles, millions or even billions of them should pass through every square centimeter of the Earth each second. For years the favorite suspect was the WIMP, a weakly interacting massive particle. No formal definition exists, but it feels gravity plus forces no stronger than the weak nuclear force. A particle of roughly 100 GeV/c2 would naturally leave behind just the right abundance of dark matter. Supersymmetric extensions of particle physics predict exactly such a particle, and the coincidence earned the name "WIMP miracle."

  Experiments attack the problem from three sides. Direct detection looks for dark matter bumping into atomic nuclei in a detector. Indirect detection looks for gamma rays, neutrinos, or antimatter from dark matter annihilating or decaying in space. Collider searches try to make the particles in the Large Hadron Collider, where they would betray themselves as missing energy.

  To escape cosmic-ray interference, direct-detection experiments sit deep underground, at laboratories such as SNOLAB in Canada, LNGS in Italy, CJPL in China, and SURF in the USA. Cryogenic detectors, chilled below 100 mK, include SuperCDMS and EDELWEISS. Liquid xenon detectors, including LZ at SURF, XENONnT at LNGS, and PandaX-4T at CJPL, have led in sensitivity for the last decade. DarkSide-20k, an argon project, is in development.

  The result so far is silence. By late 2025, no confirmed WIMP detection had been made, and the LZ team reported excluding WIMP cross-sections above 9 GeV/c2. LZ also made the first detection of boron-8 solar neutrinos in such a detector, which marks the start of a "neutrino fog" that will muddy future searches. The DAMA collaborations alone claim a yearly rise and fall in their event rate, but that remains in tension with the more sensitive experiments.

  The collider side has been equally lean. Through the end of the LHC's run 3 in 2025, the ATLAS and CMS teams had found no supersymmetric partners. They pushed limits on gluinos beyond 2.4 TeV and on charginos and neutralinos beyond 1 TeV in many scenarios. Attention has drifted toward long-lived particles, dark jets, and dark photons.

  Axions offer another path. Frank Wilczek and Steven Weinberg independently proposed them in 1978, building on a 1977 theory by Peccei and Quinn meant to solve the strong CP problem. Axion hunters use microwave cavities in strong magnetic fields to convert axions into photons. By the early 2020s, the Axion Dark Matter Experiment had reached sensitivity to the plausible DFSZ axion model.

  5051525453

• 8. Black Holes and Other Rivals

  ›

  A minority of astrophysicists argue that the real culprit is gravity, not missing matter. They propose modifying the laws of general relativity, with names such as modified Newtonian dynamics (MOND), tensor-vector-scalar gravity, and entropic gravity. General relativity is well tested within the Solar System, but its reach across galaxies is less proven. The trouble is that evidence for dark matter arrives from many independent directions at once. None of the modified theories can explain every observation, and some form of dark matter would probably still be needed. There have been scattered successes, including a 2016 test of gravitational lensing in entropic gravity and a 2020 measurement of a unique MOND effect.

  Primordial black holes offer a very different rival, one with no new particle required. Yakov Zeldovich and Igor Dmitriyevich Novikov proposed them in 1966, and Stephen Hawking independently did so in 1971. They would have formed soon after the Big Bang from extremely dense pockets of matter that collapsed on their own, with no supernova involved.

  Interest surged after LIGO detected gravitational waves in 2015. The first merger it saw involved black holes of about 30 solar masses, which are hard to explain through ordinary stellar collapse. In November 2025, the LIGO, Virgo, and KAGRA collaboration reported a candidate merger involving objects lighter than the Sun. Since no known astrophysical process makes black holes below about 1.4 solar masses, confirming such objects would strongly suggest a primordial origin.

  The James Webb Space Telescope added fuel. It found supermassive black holes, in objects such as UHZ1 and GHZ2, appearing less than 500 million years after the Big Bang and outweighing their host galaxies' stars. Microlensing data from the Subaru Telescope and Voyager 1 measurements have ruled out primordial black holes making up all dark matter in certain narrow mass windows. But those limits assume every black hole has the same mass. A 2024 review found that a broad mass distribution peaking near one solar mass could explain the entirety of dark matter.

  Critics point to a formation problem. Standard slow-roll inflation produces fluctuations far too small to trigger collapse, so exotic inflation models are needed, and the abundance they predict is exponentially sensitive to small parameter changes. Proponents answer that WIMP theories now need comparable fine-tuning after so many null results. Researchers are also exploring natural mechanisms, such as the QCD phase transition, which lowered the threshold for collapse near the solar-mass scale.

  One limit keeps the question open. No gravitational waves have been detected from beyond a redshift of 1, about 6 billion years ago. Detectors lose sensitivity to lighter collisions with distance, so the earliest half of the universe's history remains out of their reach.

  1718638919