The cosmic microwave background (CMB) is the oldest light in the Universe, emitted 380,000 years after the Big Bang. At 2.7 K, it bathes the entire sky and carries a fossil imprint of the primordial cosmos.
Imagine you could photograph the Universe as it was at age 380,000 years — cosmic infancy, long before the first galaxy, the first star, the first complex molecule. That photograph exists, and it covers the entire sky, 360° around you. It is the cosmic microwave background, the oldest observable light in the Universe.
Before that epoch, the cosmos was an opaque plasma: an ultra-hot fog of free electrons, protons and violently scattered photons. As expansion cooled everything, the temperature dropped toward 3,000 K and crossed a critical threshold: electrons snapped onto protons to form neutral hydrogen. This is the recombination (poorly named — nothing 're'-combined for the first time). The Universe suddenly became transparent — photons could travel without immediately being absorbed. That liberated light, released 13.8 billion years ago, reaches us today after a long journey through an expanding Universe that has stretched its wavelength by a factor of 1,090.
So: the CMB light departed hot (3,000 K, peaking in red-orange) and reaches us cold (2.7255 K, peaking in the microwaves near 1.9 mm wavelength). We receive about 400 photons per cm³ continuously. The spectrum is an utterly perfect blackbody — the most perfect ever measured in physics, better than 0.03 % across the entire band — signalling perfect initial thermal equilibrium.
The theoretical prediction predates the discovery: in 1948 George Gamow, Ralph Alpher and Robert Herman calculated that a hot Big Bang leaves a residual ~5 K background. They were ignored. Sixteen years later, Arno Penzias and Robert Wilson accidentally discovered this radiation while calibrating a communication antenna. Nobel Prize 1978.
The CMB has three key measurable properties.
Mean temperature. T_CMB = 2.7255 ± 0.0006 K, measured by the FIRAS spectrophotometer aboard COBE (1989-1993) with benchmark precision. This corresponds to T_0 = T_recombination / (1 + z), with z ≈ 1,090.
Blackbody spectrum. The energy intensity follows Planck's law:
B_ν(T) = (2hν³/c²) · 1/(exp(hν/kT) - 1)
COBE/FIRAS measured this spectrum between 60 and 2,900 GHz to a precision of 10⁻⁴ — the best blackbody measurement in physics.
Anisotropies. The CMB is nearly uniform, but not exactly. Three levels of departure from the mean:
• Dipole: ΔT/T ≈ 10⁻³, direction toward the constellation Leo. Doppler effect due to our proper motion at 369 km/s relative to the CMB rest frame.
• Primary anisotropies: ΔT/T ≈ 10⁻⁵, imprints of density fluctuations in the primordial Universe. These are the 'seeds' of galaxies. Discovered by COBE/DMR (1992), mapped by WMAP (2003) and Planck (2013).
• Polarization: ~5-10 % in E-mode, well measured. Primordial B-mode (signature of inflationary gravitational waves) not yet confirmed — current limit r < 0.036 (BICEP/Keck 2021).
Acoustic peaks in the angular power spectrum constrain matter density, dark-energy density, spatial geometry and the Hubble constant with unprecedented precision.
The CMB is the yardstick of modern cosmology. Planck measurements (2018) pinned down ΛCDM parameters with percent-level precision.
Age of the Universe. t₀ = 13.797 ± 0.023 billion years.
Spatial geometry. Ω_tot = 1.000 ± 0.005 → the Universe is spatially flat to high precision. This is a strong prediction of cosmic inflation.
Energy content. Ω_m = 0.315 (total matter), of which Ω_b = 0.0493 (baryons). Ω_Λ = 0.685 (dark energy). Non-baryonic dark matter thus accounts for ~26.5 %.
Hubble constant. H₀ = 67.4 ± 0.5 km/s/Mpc from the CMB — in tension with local measurements (see the Hubble Constant entry).
Spectral index. n_s = 0.965 ± 0.004 → primordial fluctuations follow a slightly red-tilted spectrum, consistent with inflation.
Observing the CMB yourself? Impossible with the naked eye or an optical telescope — it lies in microwaves. A cult trick: turn on an old analog TV not connected to an antenna, find an unassigned channel. The grey static snow you see contains about 1 % CMB photons (the rest is electronic noise). One hundredth of that snow has travelled 13.8 billion years before hitting your screen.
Four generations of missions have refined our view of the CMB.
Ground-based discovery. In 1964 at Holmdel (New Jersey), Arno Penzias and Robert Wilson aimed a 6-metre horn antenna and detected a stubborn background noise at 7.35 cm wavelength — about 3 K — that they couldn't eliminate. They first blamed pigeon droppings, cleaned everything, in vain. At the same time, Robert Dicke and Jim Peebles at Princeton were preparing an experiment to detect precisely this radiation predicted by Gamow. One phone call, two joint 1965 papers, and cosmology became an experimental science.
COBE (NASA, 1989-1993). First dedicated satellite. Measured the perfect blackbody spectrum (FIRAS) and detected primary anisotropies at 10⁻⁵ (DMR). 2006 Nobel Prize to John Mather and George Smoot.
WMAP (NASA, 2001-2010). Angular resolution 35 times better than COBE. Full-sky map in five frequencies. Measured H₀, Ω_m, n_s with unprecedented precision. Results published 2003 and 2013.
Planck (ESA, 2009-2013). Resolution 2.5 times finer than WMAP, nine frequency bands (30 to 857 GHz), better sensitivity. Set the reference cosmological parameters used today. Final data released 2018-2020.
Ground and balloon experiments. BICEP/Keck (South Pole) and SPIDER (stratospheric balloon) target primordial B-mode polarization. ACT (Chile) and SPT (Antarctica) map at high resolution for cluster and large-scale structure science.
Future. LiteBIRD (JAXA, 2029) and Simons Observatory (Chile, 2025) will hunt the inflationary signature.
The CMB is often confused with other cosmic backgrounds.
Light from the Big Bang itself. The CMB is not light emitted at the Big Bang (t = 0), but light emitted 380,000 years after. Before that era the Universe was opaque like the Sun's interior. We will never see the Big Bang in photons — we would need gravitational waves or neutrinos to push further back.
Cosmic Infrared Background (CIB). Cumulative emission from all dusty galaxies since the first star, in the far infrared (100 µm to 1 mm). Overlaps with the CMB at high frequency. Not to be confused: the CIB is the sum of individual sources; the CMB is a uniform thermal fossil radiation.
Cosmic radio / X-ray / γ-ray backgrounds. Each wavelength range has its cumulative background from discrete sources. None is a thermal relic of the Big Bang — only the CMB is.
Cosmic rays. Ultra-energetic particles (protons, nuclei, γ photons) of galactic or extragalactic origin, entirely non-thermal. Nothing to do with the CMB.
Cosmological redshift. The CMB undergoes the highest redshift ever observed (z ≈ 1,090). But redshift is a general phenomenon (galaxies, quasars, supernovae), whereas the CMB is the specific emitting object whose fossil light we study.
Man-made microwave noise. Ovens, phones, satellites emit in microwaves. CMB measurements demand highly isolated sites (South Pole, Atacama, L2 space orbits) to exclude them.
Arno Penzias and Robert Wilson, at Bell Labs in Holmdel (New Jersey) in 1964. They were calibrating a 7.35 cm horn antenna for satellite telecommunication tests and stumbled onto an unexplained background noise, everywhere in the sky, ~3 K. After suspecting pigeon droppings in the antenna, they learned that Robert Dicke's team at Princeton was preparing an experiment to detect exactly the radiation predicted by Gamow. One phone call, two joint papers in 1965. 1978 Nobel Prize in Physics to Penzias and Wilson.
Not with the naked eye or an optical telescope — it lies in microwaves, invisible. Cult trick: on an old analog TV not connected to an antenna, the grey static snow you see on an unassigned channel contains about 1 % CMB photons (the rest is internal electronic noise). One hundredth of that snow has travelled 13.8 billion years before hitting the cathode-ray tube. Modern digital sets no longer display this snow, but the experiment remains doable with an old set.
Because it provides a photograph of the Universe at age 380,000 years with extreme precision, and that photograph contains the fingerprints of every cosmological parameter. The position and amplitude of acoustic peaks in its spectrum pin down the age of the Universe, its geometry, its composition (dark matter, dark energy, baryons), the Hubble constant, and the properties of primordial fluctuations. No other cosmic object delivers so much information so tightly constrained.
Yes, very slowly. Its temperature drops by ~T/14 billion years, about 2 × 10⁻¹⁰ K per year — imperceptible. In a few tens of billions of years, the CMB will be so cooled and diluted (by dark-energy-driven accelerating expansion) that it will become undetectable. Future civilizations, if any, will no longer be able to deduce from direct observation that the Universe began with a hot Big Bang. We live in a unique cosmic observation window.