Dark night sky

If the universe is full of stars, why is the night sky dark?

Darkness seems too ordinary to need explaining. But suppose, as astronomers long did, that the universe is infinite, unchanging, and filled fairly evenly with stars. Pick any direction and follow that line outward forever. Must it not eventually meet the surface of some star? And if every sightline ends on a stellar surface, why is the sky not blazing in every direction, as bright as the face of the Sun?

That is the puzzle usually named after Olbers, though it troubled Kepler and Halley before him. It is worth taking seriously, because the darkness overhead turns out to be evidence rather than a mere absence.

Make the argument quantitative first, since a vague paradox is easy to wave away. Divide the sky into nested shells of fixed thickness centred on us. A shell twice as far has four times the area, so on a uniform distribution it holds four times as many stars — and each is four times as distant, so each arrives four times fainter. The two factors cancel exactly. Every shell, near or far, contributes the same light.

Now add up infinitely many equal contributions. Since nearer stars block more distant ones, the sum settles not at infinity but at the surface brightness of a typical star — which is still intolerable. Some premise must be false.

The reflex answer is that dust absorbs the light. Does that survive scrutiny? Absorbed energy does not vanish; it warms the absorber. Bathed in starlight from every side for eternity, dust would heat until it radiated exactly as much as it received, and then glow like the stars it was meant to hide. A screen only works if there is somewhere for the heat to go, and here there is nowhere. So the dust answer is not incomplete — it is the wrong kind of answer.

Which premise is left? Look at "infinite" alongside "unchanging". Light travels at a finite speed, so looking far away is looking far back. If the universe has a finite age, then beyond some distance there is no light yet — not light blocked, but light still in transit or never emitted. The observable universe is bounded by roughly 13.8 billion years of light travel, and the first stars did not form immediately. The sum of equal shell contributions is not infinite. It stops.

Does that alone suffice? Add that stars are not eternal either: each has finite fuel and shines for a finite span, so even the shells that count hold far less accumulated light than an eternal universe would supply. And a third, smaller effect: expanding space stretches distant light to longer wavelengths and slows its arrival rate, so each photon carries less energy and fewer arrive per second. Real, but a supporting effect rather than the resolution.

Here is the twist that makes the exercise pay. The bright sky is not absent — it is shifted out of view. Point a microwave receiver anywhere and you find a nearly uniform glow at about 2.7 kelvin: the cosmic microwave background, light from roughly 380,000 years after the beginning, when the whole universe was a hot opaque fog near 3,000 kelvin. Every sightline really does end on a glowing surface. Expansion has stretched that light by a factor of about a thousand, out of the visible and into the microwave.

So: the sky is dark in visible light because the universe is young enough, and its stars short-lived enough, that not enough starlight has accumulated — and the one all-sky glow that does exist has been redshifted clean out of our eyes' reach.

• Why the cosmic microwave background is so nearly uniform, and what its faint ripples encode.
• How the finite speed of light makes every telescope a time machine.
• Why redshift is better understood as the stretching of space than as a Doppler shift.

Every term the collection defines is gathered in the glossary.