95% of the universe escapes direct observation. What we call "dark" is in truth simply unknown matter — and perhaps the biggest open question in modern physics.
Two forces we cannot see — only their effect.
The Planck mission (ESA, 2009–2013, final analysis published 2018) reconstructed the composition of the universe with unprecedented precision from the tiny temperature fluctuations in the cosmic microwave background. The result is sobering for our self-image: everything made of atoms — stars, planets, gas, ourselves — accounts for only about 5% of the universe's energy density.
The rest is split between two components we know only indirectly, through their gravitational and cosmological effects: dark matter, which behaves gravitationally like ordinary matter but emits, absorbs, or reflects no light — and dark energy, which imparts a weak, repulsive effect on the universe and accelerates its expansion. Neither phenomenon was theoretically predicted; both were inferred from observations that contradicted the established picture: dark matter in the 1930s/1970s, from galaxies moving too fast; dark energy only in 1998, from distant supernovae that appeared too faint. The following chapters show how both trails were discovered — and how little we still know about their true nature today.
In 1933, the astronomer Fritz Zwicky noticed that galaxies in the Coma Cluster were moving far faster than the cluster's visible mass should allow — he called the missing mass "dark matter," though the scientific community at first paid little attention to the observation. The decisive, widely accepted proof came only from Vera Rubin and Kent Ford in the 1970s: they measured the rotation velocity of stars and gas in spiral galaxies such as the Andromeda Galaxy as a function of distance from the galactic center.
Schematic representation after Vera Rubin & Kent Ford, 1970s — typical orders of magnitude for a spiral galaxy.
According to the laws of Newtonian gravity and the visible mass distribution, rotation velocity should decrease with increasing distance outside the dense central region — just as the orbital velocity of the outer planets decreases with distance from the Sun. Instead, the measured velocities remained nearly constant, even far outside the luminous disk. The only explanation within known physics: galaxies are surrounded by an extended, invisible halo of matter whose gravity holds the outer stars on their orbits.
Zwicky already held the decisive clue in 1933: he called the missing mass in the Coma Cluster "dunkle Materie" (in the original German: dunkle Materie) — a term he coined himself as a German-speaking Swiss astronomer at the California Institute of Technology. His measurement method (the virial theorem applied to cluster velocities) was methodologically shaky, his numbers rough — and his idea was dismissed as a curiosity for decades. Only Rubin's precise, nearly irrefutable rotation curves turned a loner's footnote into one of the central pillars of modern cosmology.
Rotation curves were only the beginning. Since then, independent evidence has accumulated from entirely different observational methods — each on its own supporting the picture of an invisible but gravitationally active component.
In hardly any other system can dark matter be shown as vividly as in the Bullet Cluster (1E 0657-56): X-ray images from the Chandra telescope show where the hot, luminous gas mass is located — the bulk of the normal matter. Weak gravitational-lensing maps independently show where the actual total mass is located. The two distributions are clearly offset from one another.
This spatial separation can hardly be explained by modified gravity alone, and since its publication in 2006 it has stood as one of the most direct pieces of evidence for a distinct, barely interacting form of matter.
That dark matter exists is considered well established among astrophysicists. What it is remains an open question. The search focuses on three very different classes of candidates:
The most elaborate experiments search for direct detection: detectors such as XENONnT (Gran Sasso, Italy) and LUX-ZEPLIN (South Dakota, USA) observe tons of ultra-pure liquid xenon in deep underground labs, waiting for the rare signal of a WIMP particle colliding with a xenon nucleus. So far, no convincing signal has been found. Each new null result rules out further regions of the possible parameter space without answering the question — and it is honest to admit that no conclusive answer can yet be given: the nature of dark matter remains one of the most active and least resolved research fields in physics.
"Dark matter" is not a clue to what this substance is — the name merely describes what it does not do: it emits no light, absorbs none, and reflects none. It could just as well be called "invisible matter" or simply "substance X." The same is true of "dark energy": a placeholder name for a measured effect — the accelerating expansion — whose physical cause is unknown. Both terms are admissions of a gap in our knowledge, not explanations.
In 1998, two independent teams — the Supernova Cosmology Project under Saul Perlmutter and the High-Z Supernova Search Team under Brian Schmidt and Adam Riess — published an observation nobody had expected. They used Type Ia supernovae as "standard candles": these explosions of white dwarfs reach a known, nearly uniform peak brightness, which makes it possible to determine distance precisely from their apparent brightness. Yet distant — and thus, in cosmic time, earlier — supernovae appeared systematically fainter than a universe decelerated by matter and gravity would predict. The only consistent explanation: the expansion of the universe is not slowing down, it is accelerating. For this discovery, Perlmutter, Schmidt, and Riess received the 2011 Nobel Prize in Physics.
In 1917, Einstein added an extra term to his field equations, the cosmological constant Λ, to allow for a static universe — and, after Hubble's discovery of the expansion, discarded it as his "greatest blunder." After 1998, Λ returned, this time as the simplest explanation for dark energy: a constant vacuum energy density.
The problem: quantum field theory predicts a vacuum energy density about 120 orders of magnitude above the observed value — often called the worst quantitative prediction in the history of physics. Where the observed, tiny, yet non-zero value actually comes from is unknown.
If dark energy truly corresponds to a genuine cosmological constant, the future of the universe is already largely mapped out: an eternal, accelerating expansion. The favored scenario carries an unspectacular name — the Big Freeze, also called heat death. Stars burn through their fuel, galaxies recede beyond each other's observable horizon, matter decays over unimaginable timescales, until the universe reaches a cold, extremely diluted equilibrium.
Schematic evolution of the scale factor — decelerated expansion during matter domination, transition around 5 billion years ago, acceleration ever since.
Alternative scenarios — a renewed collapse (Big Crunch, given sufficient matter and gravity) or a tearing apart of spacetime itself (Big Rip, given steadily growing dark energy) — are considered far less likely based on current measurements. What continued expansion means for the edge of the observable universe and for questions beyond our physical reach is explored further in the separate overview of cosmological limits.
The path to the Big Freeze is not a matter of millions of years, but of orders of magnitude that make even cosmological timescales look small. In the Stelliferous Era, which we are currently in, stars are still burning (up to ~10¹⁴ years). After that comes the Degenerate Era, in which only white and brown dwarfs remain, followed by the Black Hole Era, in which even black holes evaporate via Hawking radiation (~10¹⁰⁰ years). What remains after that is a matter for the overview of cosmological limits — here, the short outlook suffices: if Λ stays constant, the end of the universe will be not a bang, but a fade.
Despite all the progress, at least one solid contradiction remains: two independent, high-precision methods for measuring the universe's current expansion rate (the Hubble constant H₀) yield values that do not statistically overlap.
Local measurements via the cosmic distance ladder yield about 73.0 km/s/Mpc. The back-calculation from the cosmic microwave background under the ΛCDM standard model yields about 67.4 km/s/Mpc — a difference of roughly 8–9%, which is not measurement noise. Whether the solution lies in undiscovered systematic errors, in new physics of dark energy, or in as-yet-unknown particles is one of the most active open questions in cosmology.
The Hubble tension is not a marginal problem — it concerns the fundamental quantity against which every distance in the universe is calibrated. If it dissolves as a measurement error, ΛCDM remains intact. If it is confirmed further, it could be the first concrete hint that dark energy — or dark matter — is more complex than the simple cosmological constant assumes. Either possibility makes the Hubble tension one of the most exciting open threads in this chapter of physics.