What is the cosmic microwave background?

On the familiar oval map of the early Universe, red and blue patches seem to show blazing clouds and cold hollows. In reality, almost all this sky has the same temperature. Why is the map so colourful?

The cosmic microwave background (CMB) is ancient thermal light filling the Universe. Today its spectrum corresponds to a temperature of about 2.725 K. Colour maps usually amplify tiny deviations from the average rather than show true visible colours. PDG: spectrum and temperature anisotropies, §§29.1–29.3.

Light already existed, but could not travel far

Photons existed in the hot early environment long before it became transparent. Free electrons frequently scattered them, changing their directions. This resembles light in thick fog: light is present, but a direct view of distant objects cannot get through. The cosmic “fog” was not made of water droplets; the analogy explains scattering alone.

As the medium cooled, nuclei and electrons began forming neutral atoms. The abundance of free electrons dropped sharply, and light could travel much farther without scattering. The principal epoch from which the observed primary CMB originates corresponds to roughly 380,000 years in standard cosmic history. This was a gradual transition, not a switch flicked everywhere at once. ESA: the early medium and freely travelling light, PDG: last scattering, §29.3.3.

So the CMB should not be explained merely as new photons born when atoms captured electrons. The main change was that existing thermal radiation ceased to be tightly coupled to the plasma. Since then, cosmic expansion has stretched its wavelengths and lowered its temperature to today’s microwave background.

A spectrum's temperature and a map's patches are measured differently

One experiment asks: how much radiation arrives at different frequencies? The FIRAS instrument aboard COBE compared sky signals with a calibrated thermal reference. The measured spectrum is extraordinarily close to that of a blackbody: an ideal thermal emitter. That is how we obtain a temperature, rather than simply calling faint light “cold”. NASA/LAMBDA: how FIRAS worked.

A different question is: is the temperature identical in different directions? WMAP and Planck maps reveal small differences. To see these, researchers subtract the mean background, account for the large effect of our motion, and separate Galactic emission. Dust and the CMB, for example, have different frequency dependences; observing several frequencies helps separate components. This is an analysis with noise and uncertainties, not a straightforward photograph of a colourful sky. Planck: component-separation methods.

Image contrast does not heat the sky

Imagine three regions with arbitrarily chosen deviations of −100, 0 and +100 microkelvin from a reference temperature of 2.7255 K. Show them with two palettes. The colours differ more strongly in the second row, while the numbers stay unchanged.

The same deviations, different colour contrast

Low contrast

−100 µK

0 µK

+100 µK

Enhanced contrast

−100 µK

0 µK

+100 µK

Illustrative region 1

Deviation ΔT: −100 µK
Total temperature: 2.7254 K

Illustrative region 2

Deviation ΔT: 0 µK
Total temperature: 2.7255 K

Illustrative region 3

Deviation ΔT: +100 µK
Total temperature: 2.7256 K

Three invented values to explain a palette: not a Planck map, measured pixels or the shape of the Universe. Both rows use the same order: −100, 0, +100 µK relative to T₀ = 2.7255 K. The numbers do not change. Colour encodes the sign of a deviation; patch areas are schematic, not angular sizes. 1 µK = 0.000001 K.

Increasing the contrast and printing the map more vividly cannot increase the physical temperature difference. This makes the legend, units and processing description as necessary as the image itself. On the published Planck map, warmer and cooler directions are likewise represented by assigned colours. Our diagram explains that principle without reproducing observed pixels. IPAC/Planck: what map colours mean.

Branch · Numerical exampleOptional · you can skip this

How small is 100 microkelvin?

A microkelvin is a millionth of a kelvin. For the illustrative deviation in our diagram,

$100\ \mu\mathrm K=10^{-4}\ \mathrm K.$

Let $T_0=2.7255\ \mathrm K$ be the reference mean. The relative deviation is then

\begin{aligned} \frac{|\Delta T|}{T_0}&=\frac{0.0001}{2.7255}\\ &\approx3.7\times10^{-5}. \end{aligned}

That is approximately 0.0037%. The complete temperatures are 2.7254, 2.7255 and 2.7256 K. They are all close, although the palette deliberately makes the differences visible. Our ±100 µK example is neither a limit on every real patch, an instrument’s precision, nor the uncertainty in the mean temperature. PDG: mean level, dipole and small anisotropies distinguished.

What do we read in this ancient light?

The patches are related to early perturbations in matter and radiation. But “a red patch simply means more matter right there” is too crude. Observed temperature combines intrinsic temperature differences, motion of the medium, and gravity’s influence on light. Later structures also alter photon paths, including through gravitational lensing. PDG: temperature differences and lensing, §§29.5.1–29.5.4.

Cosmologists compare the characteristic angular scales and correlations of these differences, as well as polarization, with model calculations. This constrains the Universe’s composition and geometry. It is not a direct photograph of dark-matter particles or separate proof of every assumption in a model.

“Oldest light” refers to our electromagnetic view of an early transparent epoch. It is not a photograph of time zero. Some photons scattered again later; not every sky signal followed an unchanged path. Earlier epochs are also studied through indirect traces, rather than an ordinary photograph through opaque plasma.

Start a colourful map by reading its legend: what was subtracted, what units describe the residuals, and how were colours assigned? Astronomy beyond visible light explains why an instrument can detect what the eye cannot.