At a glance
In the standard cosmological model, ΛCDM, everything we can see — stars, gas, planets, people — makes up only about one-twentieth of the Universe’s present energy density. Roughly a quarter is dark matter, which gravitates but neither emits nor absorbs light, and about two-thirds is dark energy, which drives the accelerating expansion of space. This page focuses on these two dark components; Big Bang basics are covered in a separate cosmology dossier.
| Component / parameter | Measured value (68% uncertainty) | Approximate share of today’s energy density | Notes |
|---|---|---|---|
| Ordinary (baryonic) matter | Ωbh2 = 0.02237 ± 0.00015 | ≈ 4.9% (Ωb ≈ 0.049) | Atoms of all kinds |
| Cold dark matter | Ωch2 = 0.1200 ± 0.0012 | ≈ 26.5% (Ωc ≈ 0.265) | Nature unknown |
| Total matter | Ωm = 0.3153 ± 0.0073 | ≈ 31.5% | Includes a small neutrino term |
| Dark energy (cosmological constant) | ΩΛ = 0.6847 ± 0.0073 | ≈ 68.5% | 0.6889 ± 0.0056 with BAO data |
| Radiation (CMB photons) | Ωγ = (5.38 ± 0.15) × 10−5 | ≈ 0.005% | Negligible today; dominant in the very early Universe |
| Hubble constant | H0 = 67.36 ± 0.54 km s−1 Mpc−1 | — | 67.66 ± 0.42 with BAO; see “Hubble tension” below |
| Age of the Universe | 13.797 ± 0.023 billion years | — | Model-dependent (assumes ΛCDM) |
| DESI DR2 (March 2025): dark-energy equation of state w0wa | Preferred region w0 > −1, wa < 0 | — | Preferred over ΛCDM at 3.1σ (BAO + CMB); 2.8–4.2σ adding supernovae, depending on sample |
Planck values: TT,TE,EE+lowE+lensing column of Table 2 in Planck 2018 VI (published 2020). Ωb and Ωc are derived here by dividing by h2 with h = 0.6736 and are rounded. Ωγ: Particle Data Group 2024. DESI significances as stated in the DR2 abstract; none reaches the conventional 5σ discovery threshold. Checked October 2026.
How we know dark matter exists
The first hint came in 1933, when Fritz Zwicky measured the velocities of galaxies in the Coma Cluster. Applying the virial theorem, he found the galaxies move too fast to be bound by their visible stars, and spoke of “dunkle Materie” — dark matter. In the 1970s, Vera Rubin and Kent Ford measured the rotation of the Andromeda galaxy and then of many other spirals, and found that orbital speeds stay roughly constant far beyond the visible disc instead of falling off. These flat rotation curves became a cornerstone of the case.
Gravitational lensing weighs mass directly, whatever form it takes, by measuring how it bends the light of background galaxies. The most striking case is the merging cluster 1E 0657−558, the “Bullet Cluster”, at redshift 0.296. X-rays show the hot gas — most of the ordinary matter — slowed in the middle of the collision, while weak lensing shows most of the mass has passed through with the galaxies. In 2006 Clowe and colleagues found that the centre of total mass is offset from the peak of baryonic mass at 8σ significance, which is difficult to explain by modifying gravity alone.
The most precise evidence comes from the cosmic microwave background (CMB). Before recombination, photons and baryons oscillated as one fluid in dark-matter potential wells, imprinting acoustic peaks in the CMB power spectrum; the peak heights separately fix the baryon density and the density of matter that does not interact with light. Planck’s fit gives a dark-matter density about five times the baryon density.
Interactive lab: the rotation curve and the unseen mass
The lab opens on a Milky Way-like model whose dark-matter halo keeps the curve nearly flat, giving about 232 km/s at the Sun’s distance against a measured 229 km/s. Choose “Visible matter only” and the speed there drops to about 168 km/s, then keeps falling beyond the bright disc, far below the measured points (white dots). Back on the “Milky Way-like model”, notice that about three-quarters of the mass inside 25 kpc has to be unseen.
What could dark matter be?
Dark matter must be stable over cosmic time, interact very weakly (if at all) with light and ordinary matter, and be “cold” enough not to erase small-scale structure. No Standard Model particle fits — ordinary neutrinos are too light and fast — so most candidates involve new physics.
WIMPs (weakly interacting massive particles) became popular because particles with roughly electroweak-scale interactions, produced thermally in the hot early Universe, would naturally freeze out at about the right abundance — the so-called “WIMP miracle”. The Particle Data Group review notes that this coincidence has no fundamental link to the weak scale. Axions, very light bosons originally proposed to solve the strong-CP problem of quantum chromodynamics, are an equally compelling candidate and are searched for with resonant-cavity experiments such as ADMX.
Sterile neutrinos with keV-scale masses could decay and produce an X-ray line. A tentative line near 3.5 keV was reported, but non-detections in many galaxies, groups and dwarf galaxies — notably Draco — cast strong doubt on a dark-matter origin. Primordial black holes, formed from collapsing density fluctuations in the early Universe, are not particles at all; microlensing and other observations constrain the fraction of dark matter they can make up over many mass ranges, and whether any window remains fully open is still debated.
Sources: [6]
The hunt: direct, indirect and collider searches
Direct detection experiments wait for a halo particle to hit an atomic nucleus in an ultra-quiet underground detector. LUX-ZEPLIN (LZ), with 10 tonnes of liquid xenon in South Dakota, reported in 2024 a 4.2 tonne-year (280 live days) null result, setting a 90% confidence limit on the spin-independent WIMP–nucleon cross-section of 2.2 × 10−48 cm2 for a 40 GeV/c2 WIMP (Physical Review Letters, 2025).
In December 2025 LZ presented an analysis of 417 live days collected between March 2023 and April 2025. Again no WIMPs were seen; the search was extended for the first time down to 3 GeV/c2 and is world-leading above 5 GeV/c2. LZ also detected nuclear recoils from boron-8 solar neutrinos at 4.5σ, a background that future searches cannot avoid. LZ aims for over 1,000 live days by 2028; a successor, XLZD, is being designed. XENONnT (Gran Sasso, Italy) published a 3.1 tonne-year null result in 2025, with a strongest limit of 1.7 × 10−47 cm2 at 30 GeV/c2.
Indirect detection seeks annihilation or decay products. The Fermi Large Area Telescope has set strong limits from dwarf spheroidal galaxies; a gamma-ray excess from the inner Galaxy fits a WIMP of a few tens of GeV, but unresolved millisecond pulsars could explain it, and it remains inconclusive. Collider searches at the Large Hadron Collider look for missing transverse momentum and new resonances; none has been found, and collider data alone could not prove a particle is the cosmological dark matter.
Modified gravity: MOND and its challenges
An alternative view holds that there is no missing mass, and that our law of gravity fails at very low accelerations. Modified Newtonian Dynamics (MOND), proposed by Mordehai Milgrom in the early 1980s, changes the dynamics below a characteristic acceleration scale. It describes the rotation curves of many individual galaxies using only their visible matter with notable economy, and it reproduces the tight empirical relation between a galaxy’s baryonic mass and its rotation speed.
The difficulties appear on larger scales. Galaxy clusters still require unseen mass under MOND, and merging clusters such as the Bullet Cluster show lensing mass separated from the ordinary matter. Theories without dark matter also struggle with the CMB peaks and the matter power spectrum, which ΛCDM fits naturally. The near-simultaneous arrival of gravitational waves and light from the neutron-star merger GW170817 also ruled out modified theories in which the two travel at different speeds. Most cosmologists therefore favour particle dark matter, while ΛCDM itself faces small-scale puzzles such as the diversity of rotation curves.
Interactive lab: weighing mass with lensing
The lab opens on a galaxy-cluster lens of about 3 × 1014 solar masses, whose Einstein ring comes out at nearly 48 arcseconds in radius. Change the lens mass and watch the ring grow or shrink with its square root: this is how lensing weighs mass, whatever form it takes. Then pick the microlensing preset: a single star makes a ring only about half a milliarcsecond in radius, yet brightens the background star more than threefold over a few weeks.
Dark energy: from discovery to an evolving question
In 1998 two teams — the Supernova Cosmology Project and the High-z Supernova Search Team — announced that distant Type Ia supernovae appear fainter than expected in a decelerating Universe. Both concluded that cosmic expansion is accelerating. Saul Perlmutter, Brian Schmidt and Adam Riess received the 2011 Nobel Prize in Physics for the discovery.
The simplest description is Einstein’s cosmological constant, Λ, equivalent to a vacuum energy with equation-of-state parameter w = −1 that never changes. It fits most data, but naive quantum estimates of vacuum energy exceed the observed value by many orders of magnitude. Alternatives include evolving scalar fields and modified gravity. Surveys test them by letting w vary, usually as w(a) = w0 + wa(1 − a), where a is the cosmic scale factor; ΛCDM corresponds to w0 = −1, wa = 0.
The Dark Energy Spectroscopic Instrument (DESI) maps baryon acoustic oscillations, a standard ruler of about 147 Mpc, in millions of galaxies and quasars. Its second data release analysis (DR2, March 2025) found that DESI BAO combined with CMB data prefer w0 > −1 and wa < 0 over ΛCDM at 3.1σ, and at 2.8–4.2σ when supernovae are added, depending on which supernova sample is used. This is suggestive but not a discovery: a later Dark Energy Survey reanalysis with an updated supernova calibration lowered the significance for the DES supernova combination from 4.2σ to 3.2σ, corresponding to only weak Bayesian preference (odds of about 5:1). Whether dark energy truly evolves remains open as of October 2026.
The Hubble tension
The present expansion rate, H0, can be measured locally with a distance ladder — parallaxes, Cepheid variable stars and Type Ia supernovae — or inferred from the early Universe by fitting ΛCDM to the CMB. The SH0ES team obtained H0 = 73.04 ± 1.04 km s−1 Mpc−1 in 2022, while Planck 2018 gives 67.36 ± 0.54 km s−1 Mpc−1. The SH0ES authors describe the gap as a 5σ difference.
Debate centres on two possibilities: unrecognised systematic errors, for example in distance-ladder calibration, or an incomplete ΛCDM, such as new physics changing the sound horizon before recombination. No explanation has achieved consensus; independent methods, including gravitational-wave “standard sirens”, are being developed.
The next decade of surveys
New wide surveys aim to measure expansion history and structure growth precisely enough to separate a cosmological constant from evolving dark energy. ESA’s Euclid mission released its first “Quick” data (Q1, 19 March 2025): 63.1 square degrees of deep fields with about 26 million galaxies. According to ESA’s current timeline, a “DR1-Foundation” release of raw and calibrated images covering about 1,900 square degrees is planned for November 2026, with the complete DR1, including higher-level weak-lensing and clustering products, in mid-2027.
The NSF–DOE Vera C. Rubin Observatory in Chile started its ten-year Legacy Survey of Space and Time (LSST), as announced at the end of June 2026, adding weak lensing, clustering and supernova data. NASA’s Nancy Grace Roman Space Telescope launched on 30 August 2026 aboard a Falcon Heavy from Kennedy Space Center. It is travelling to the Sun–Earth L2 point, a journey of about three months, and NASA expects its first images by early 2027.
Going further: public data and tools
Most data behind this page are public. The Planck Legacy Archive (ESA) holds Planck maps, power spectra, likelihoods and parameter chains. NASA’s LAMBDA (Legacy Archive for Microwave Background Data Analysis) gathers data and software from many CMB experiments.
DESI’s public releases include the Early Data Release (June 2023) and Data Release 1 (March 2025, 18.7 million new spectra); the cosmology chains for the DR2 BAO analysis are also published there. The Sloan Digital Sky Survey (SDSS) released DR20 in July 2026.
A good first exercise is to compute a CMB power spectrum with an open Boltzmann code such as CAMB or CLASS and compare ΛCDM with a w0wa model.
Known vs open questions
Well established
- Gravitational evidence for dark matter is consistent across galaxies, clusters, lensing and the CMB.
- Dark matter is mostly non-baryonic and roughly five times as abundant as ordinary matter.
- Cosmic expansion is accelerating; ΛCDM fits most cosmological data with a handful of parameters.
- Direct searches have excluded large regions of WIMP parameter space without detecting a signal.
Open
- What dark matter is made of — WIMPs, axions, something else, or a mix.
- Whether dark energy is a cosmological constant or evolves with time (DESI hints at 2.8–4.2σ, not yet a discovery).
- The origin of the Hubble tension: systematics or new physics.
- Small-scale puzzles in galaxies, and whether any modified-gravity theory can match all the data.
Sources and further reading
- Planck Collaboration — Planck 2018 results. VI. Cosmological parameters (Astronomy & Astrophysics 641, A6, 2020) — EDP Sciences / ESA
- Particle Data Group — Astrophysical Constants and Parameters (Review of Particle Physics, 2024) — Particle Data Group / Physical Review D
- DESI Collaboration — DESI DR2 Results II: Measurements of Baryon Acoustic Oscillations and Cosmological Constraints (2025) — DESI / arXiv
- Bertone, G. & Hooper, D. — A History of Dark Matter (Reviews of Modern Physics 90, 045002, 2018) — APS / arXiv
- Clowe, D. et al. — A direct empirical proof of the existence of dark matter (The Astrophysical Journal Letters, 2006) — AAS / arXiv
- Baudis, L. & Profumo, S. — Dark Matter (Review of Particle Physics, Particle Data Group, 2024) — Particle Data Group / Physical Review D
- LZ Collaboration — Dark Matter Search Results from 4.2 Tonne-Years of Exposure of the LUX-ZEPLIN (LZ) Experiment (Physical Review Letters 135, 011802, 2025) — APS / arXiv
- Berkeley Lab News Center — LZ sets a world’s best in the hunt for galactic dark matter (December 2025) — Lawrence Berkeley National Laboratory
- XENON Collaboration — WIMP Dark Matter Search using a 3.1 Tonne-Year Exposure of the XENONnT Experiment (Physical Review Letters 135, 221003, 2025) — APS / arXiv
- Nobel Prize — The Nobel Prize in Physics 2011: press release — Royal Swedish Academy of Sciences
- DES Collaboration — The Dark Energy Survey Supernova Program: A Reanalysis of Cosmology Results and Evidence for Evolving Dark Energy with an Updated Type Ia Supernova Calibration (2025) — DES / arXiv
- Riess, A. G. et al. — A Comprehensive Measurement of the Local Value of the Hubble Constant with 1 km/s/Mpc Uncertainty (The Astrophysical Journal Letters, 2022) — AAS / arXiv
- ESA Cosmos — Euclid DR1 timeline — European Space Agency
- NASA — NASA’s Dark Universe-Seeking Nancy Grace Roman Space Telescope Launches (August 2026) — NASA