CMB Dipole: How Our Motion Through the Universe Warms One Side of the Sky

What the CMB dipole is: a speedometer in the sky

The cosmic microwave background is, to a first approximation, a perfect blackbody at T₀ = 2.72548 K filling the sky uniformly in every direction. But that uniformity is only exact in one special frame — the CMB rest frame, in which the radiation has no net momentum. Any observer moving relative to that frame sees the sky as slightly hotter in the direction of motion and slightly cooler behind, producing a smooth cos θ pattern from one pole to the other.

This pattern is the dipole anisotropy, the ℓ=1 spherical-harmonic component of the temperature map. Its measured amplitude is 3.3621 ± 0.0010 mK — about 100 times larger than the primordial fluctuations that seeded galaxies. Because it arises purely from our motion, the dipole is not a relic of the early universe; a hypothetical observer at rest in the CMB frame would see no dipole at all. It is the one anisotropy we could, in principle, transform away by changing our velocity.

The mechanism: relativistic Doppler boosting of a blackbody

An observer moving with velocity v through an isotropic radiation field sees each photon Doppler-shifted according to its angle θ from the direction of motion. For a blackbody, a Doppler shift is equivalent to a temperature change, so the observed temperature becomes direction-dependent:

• Exact form: T(θ) = T₀ · √(1 − β²) / (1 − β·cos θ), where β = v/c.
• First order (β ≪ 1): ΔT/T₀ ≈ β·cos θ — a pure dipole.
• Second order: the expansion leaves a residual (β²/2)·cos 2θ term, a kinematic quadrupole of order (v/c)² ≈ 1.5×10⁻⁶.

The crucial point is that the shifted spectrum remains a blackbody in every direction — only its temperature changes. This is why the dipole is described as a temperature dipole rather than a distortion of the Planck spectrum. Measuring the amplitude ΔT and dividing by T₀ directly yields β, and hence our speed: v = c·(ΔT/T₀) = c × (3.3621 mK / 2725.48 mK) ≈ 370 km/s.

The numbers: our velocity and where we're headed

Planck's 2018 analysis pins the solar system's velocity relative to the CMB at v = 369.82 ± 0.11 km/s, directed toward Galactic coordinates l = 264.021°, b = 48.253° — a point in the constellation Crater, near the Leo–Hydra border. A worked check: β = v/c = 369.82 / 299792 = 1.233×10⁻³, and β·T₀ = 1.233×10⁻³ × 2725.48 mK = 3.36 mK, matching the observed amplitude.

This solar velocity is a sum of several motions. Removing the Sun's orbit around the Galaxy (~220 km/s toward Cygnus) and the Galaxy's motion within the Local Group yields the velocity of the Local Group of galaxies itself: about 620–627 km/s toward l ≈ 276°, b ≈ 30°. That bulk flow points roughly toward the Great Attractor and the Shapley Supercluster, and away from a large underdense region dubbed the Dipole Repeller.

How it is observed and separated from the sky

The dipole was first detected in the late 1960s and 1970s from balloons and U-2 aircraft, then measured precisely by satellites. COBE (1990s) fixed the amplitude near 3.35 mK; WMAP refined the direction; and Planck (2013–2018) delivered the current 0.03% precision. Because the dipole is so large, missions actually use it as a calibration source — its known cos θ shape and time-varying orbital modulation provide an on-sky standard candle for gain calibration.

• Orbital dipole: Earth's ~30 km/s orbit around the Sun adds a small, precisely predictable annual modulation used to cross-check calibration.
• Foreground masking: the Galactic plane is masked; the dipole is fit over the clean high-latitude sky.
• Frequency independence: because the shift preserves a blackbody, the dipole has the same ΔT at all frequencies, distinguishing it from foregrounds like synchrotron or dust.

Cousins and contrasts: dipole vs. primordial anisotropy

It is essential to distinguish the kinematic dipole from the primordial anisotropies that dominate at ℓ ≥ 2. The primordial fluctuations (the acoustic peaks near ℓ≈220, the Sachs–Wolfe plateau) are frozen features of the universe at recombination, ~380,000 years after the Big Bang, and are the same for all observers up to their tiny motion-induced corrections. The dipole is observer-dependent and would vanish in the right frame.

• vs. cosmological redshift: redshift stretches wavelengths uniformly with expansion; the dipole is a directional blueshift/redshift from local peculiar velocity, not expansion.
• vs. the intrinsic (Sachs–Wolfe) dipole: theory predicts a genuine primordial ℓ=1 term exists too, but it is buried far beneath the ~1000×-larger kinematic signal and cannot yet be isolated.
• vs. aberration and modulation: our motion also aberrates and modulates the higher-ℓ pattern at the β level — a small but detected effect Planck measured to confirm the dipole is kinematic.

Significance and the open 'dipole tension'

The CMB dipole defines the cosmic rest frame — the natural reference against which all peculiar velocities and bulk flows are measured. It anchors the standard cosmological model's assumption that, viewed from the right frame, the universe is isotropic.

That assumption is now under scrutiny. If the dipole is purely kinematic, then any distant, isotropically distributed population — radio galaxies, quasars — should show a matching dipole in their number counts, of the same amplitude and direction (the Ellis–Baldwin test). Yet several studies of NVSS radio sources and, notably, ~1.4 million WISE quasars (Secrest et al., 2021) find a number-count dipole aligned with the CMB direction but roughly twice the expected amplitude — a >4σ discrepancy. Explanations range from local structure and survey systematics to a genuine violation of the Cosmological Principle. Whether the 'dipole tension' is a measurement artifact or a crack in the standard model remains one of the live debates in modern cosmology.

CMB dipole compared to other CMB anisotropies and reference velocities

Feature	Angular scale (ℓ)	Amplitude (ΔT)	Physical origin
Kinematic dipole	ℓ = 1	3.3621 mK	Solar system peculiar velocity (Doppler)
Primordial monopole	ℓ = 0	2.72548 K (mean)	Blackbody temperature of the CMB
Acoustic peak fluctuations	ℓ ≈ 200–1000	~10⁻⁵ K (~70 µK)	Baryon–photon oscillations at recombination
Sachs–Wolfe plateau	ℓ ≲ 30	~30 µK	Gravitational potential at last scattering
Kinematic quadrupole	ℓ = 2	~1.2 µK	Second-order (v/c)² boosting term