Dark Matter: Evidence, Theories, and Open Questions
Why physicists are confident something invisible is out there — rotation curves, lensing, the CMB — and why nobody knows what it actually is.
By Velloreth
There is a problem at the center of modern physics, and it is not subtle. About 85% of the matter in the universe does not interact with light. It does not emit, absorb, or reflect electromagnetic radiation. Every telescope, in every wavelength, sees the other 15%. Yet the evidence that this invisible mass exists is among the most robust in all of science — drawn from a dozen independent lines of observation, each with its own systematic errors and assumptions, all pointing the same way.
Nobody knows what it is. That is the open question this post is about.
The evidence, one line at a time
Dark matter is a hypothesis built to explain observations that otherwise require either unseen mass or a breakdown of gravity. The strongest case comes from several independent phenomena that would each need their own ad hoc fix.
Galaxy rotation curves
In the 1930s and 1940s, Fritz Zwicky measured the velocities of galaxies in the Coma Cluster and found them moving too fast for the cluster’s visible mass to hold them gravitationally. He inferred unseen mass — he called it dunkle Materie — but the idea did not gain traction until the 1970s, when Vera Rubin and Kent Ford measured rotation curves of spiral galaxies with precision.
The prediction from Newtonian gravity is straightforward: outside the visible disk, orbital velocity should fall off as 1/√r, the same way planetary velocities drop with distance from the Sun. Instead, rotation curves stay flat — stars at the galaxy’s edge orbit as fast as those much closer in. The only way to keep velocities constant with radius is a mass distribution that grows linearly with radius: a diffuse halo of unseen matter extending far beyond the visible galaxy.
This is not a subtle effect. Our own galaxy rotates at roughly 240 km/s at the Sun’s location, with little change out to the largest measurable radii. The implied mass is several times the visible mass.
Galaxy clusters and the Bullet Cluster
Clusters of galaxies are held together by their total mass, which can be estimated three independent ways: the velocities of member galaxies, the temperature and pressure of the hot X-ray-emitting gas between them, and the gravitational lensing of background objects.
All three methods agree that most of the mass is not in the visible gas or stars. The Bullet Cluster (1E 0657-558) makes the case especially cleanly: two clusters collided, and the hot gas (most of the ordinary mass) was shocked and slowed, while the galaxies passed through. Lensing maps show the bulk of the mass traveled with the galaxies, not the gas. If the missing mass were ordinary matter somehow hidden, it should have stuck with the gas. Instead, it behaved like something that does not interact except through gravity.
Gravitational lensing
Einstein’s general relativity predicts mass bends light. Massive clusters act as lenses, distorting and magnifying background galaxies into arcs — strong lensing — or producing subtle coherent distortions in large samples — weak lensing. Both methods produce mass maps independent of any assumption about what the mass is made of, and the maps show far more mass than the light accounts for.
The cosmic microwave background
The CMB is the afterglow of the Big Bang, and its temperature fluctuations — anisotropy of about one part in 100,000 — encode the composition of the early universe. The precise pattern of peaks in the CMB power spectrum, measured by COBE, WMAP, and especially Planck, fits the Lambda-CDM model with three parameters for the energy budget: ordinary matter, dark matter, and dark energy.
Ordinary matter alone cannot produce the observed peak structure. Baryons interact with radiation before recombination; that coupling sets up acoustic oscillations whose pattern depends on the baryon density. Dark matter, which does not interact with radiation, gravitates from the start and provides the potential wells into which ordinary matter later falls. The CMB fits best with roughly 25% of the total energy density in dark matter, about 5% ordinary, and 70% dark energy — and this composition independently predicts the abundances of light elements from Big Bang nucleosynthesis and the timeline of galaxy formation.
Structure formation
Simulations of cosmic structure growth with only ordinary matter produce a universe that looks wrong: too few small-scale structures, structures forming too late. With cold dark matter added, simulations reproduce the observed cosmic web — filaments, voids, and halos — from the initial conditions measured in the CMB. Galaxies exist at redshift z ~ 10, which requires gravitational collapse to have started early, which requires the non-baryonic component that decoupled from radiation and began clumping before recombination.
No single line of evidence is airtight alone. Together, they are. The standard summary: dark matter constitutes about 25% of the universe’s total energy density and over 80% of its matter.
What dark matter is not
The evidence tells us what dark matter does — gravitates, does not interact electromagnetically, does not self-interact strongly, was non-relativistic in the early universe — but not what it is. Two broad categories of explanation have been tested.
The particle route
The Standard Model of particle physics contains no suitable candidate. Every known particle is either charged, interacts too strongly, or is massless or nearly so. So dark matter is, by most assumptions, a new particle — stable on cosmological timescales, neutral, and produced with the right relic abundance in the early universe.
The three leading candidates:
WIMPs — weakly interacting massive particles, masses roughly 10 GeV to a few TeV, interacting via the weak force. They arise naturally in supersymmetric extensions of the Standard Model, and thermal freeze-out in the early universe produces roughly the right abundance for weak-scale interactions — the famous “WIMP miracle.” Direct detection experiments (LUX-ZEPLIN, XENONnT, PandaX) look for nuclear recoils from WIMP scattering; indirect searches look for annihilation products in dense regions; colliders try to produce them.
Decades of searching have found nothing. LZ’s 2025 results pushed exclusion limits into the region where solar neutrinos create an irreducible background — the “neutrino fog” — meaning further progress requires discriminating WIMP signals from neutrino scattering, not just building bigger detectors. The LHC has also failed to find supersymmetric particles, constraining the natural parameter space. WIMPs are not dead, but the simplest versions have been under pressure for years.
Axions — ultralight particles originally proposed to solve the strong CP problem in quantum chromodynamics. They are produced non-thermally and behave as cold dark matter despite tiny masses (micro-eV range plausible). Experiments like ADMX use microwave cavities in strong magnetic fields to convert axions to detectable photons. Axion searches have expanded dramatically as WIMP results came up empty; the parameter space is vast, but experimental sensitivity is finally reaching motivated ranges.
Primordial black holes — black holes formed from density fluctuations in the early universe, not from stellar collapse. They interact only gravitationally, need no new particles, and LIGO’s detections of mergers between black holes of unexpectedly large mass revived interest. Constraints from microlensing, CMB distortions, and Hawking radiation narrow the allowed mass windows significantly; PBHs can be at most part of the dark matter except in specific gaps.
The gravity route
Modified Newtonian Dynamics (MOND), proposed by Milgrom in 1983, modifies the law of inertia below a critical acceleration (~1.2 × 10⁻¹⁰ m/s²) and fits galaxy rotation curves with remarkable precision — often better than dark matter halos do for individual galaxies, using one free parameter per galaxy instead of a fitted halo profile.
The problem is scope. MOND has no single relativistic theory that reproduces all its successes; TeVeS and other relativistic extensions have been constrained by gravitational wave observations (GW170817 constrained the speed of gravitational waves to essentially the speed of light, ruling out several MOND-inspired theories). More decisively, the CMB power spectrum is very hard to reproduce without a non-baryonic component; modified gravity alone does not fit. The Bullet Cluster’s mass-galaxy coincidence is also awkward for purely modified gravity. Most of the astrophysics community concludes that some form of dark matter exists, even if gravity also needs revision at some level.
The open questions
What keeps thousands of physicists working on this:
- What is the particle (if it is a particle)? WIMP parameter space is compressed but not closed; axion searches are scaling up; hidden-sector and ultralight candidates remain open.
- Is there dark matter at small scales? Lambda-CDM predicts cuspy halo cores and abundant satellite galaxies; observations often show cored profiles and fewer satellites — the “core-cusp” and “missing satellites” problems. Baryonic physics (supernova feedback, tidal effects) may resolve them; warm dark matter is another option.
- Are dark matter and dark energy related? Some theories unify them as different aspects of one field; no evidence yet forces this.
- How does dark matter interact with itself, if at all? Strong self-interaction is constrained by the Bullet Cluster; weak self-interaction remains possible and would affect halo structure.
Where the searches go next
The experimental program is diversifying. Direct detection is entering the neutrino fog, where next-generation detectors need directional sensitivity or new materials to push further. Axion experiments are scaling to cover motivated mass ranges within the decade. Gravitational-wave observatories and 21-cm cosmology — mapping neutral hydrogen across cosmic time — will probe structure formation in ways independent of light. And quantum sensors, originally developed for other purposes, are being adapted for ultralight dark matter detection.
If a signal appears in any of these, it will be one of the most consequential discoveries in physics: a new particle, a new interaction, or evidence that gravity is not what we thought. Until then, the rotation curves keep coming out flat, the CMB keeps fitting Lambda-CDM, and the universe keeps about 85% of its matter hidden from us — present in every measurement, absent from every telescope.