Home / Learn / Science / Cosmology A to Z / Lesson 11.1
All lessons

Module 11: Inflation and the first instant · Lesson 11.1 · about 16 min

The horizon and flatness problems

Why is the CMB the same temperature in places that never met? Why is space so close to flat? Three puzzles the plain Big Bang cannot explain, and the numbers behind them.

Key points, meanings and formulas on one page: choose "Save as PDF" in the print box.

Module 11 · about 16 minutes. Read first: 5.9 Horizons and 5.4 The shape of space. The Big Bang theory explains the expanding universe, the CMB and the light elements beautifully. But it takes the starting state for granted, and that starting state is very odd. By the end you will be able to explain, with numbers, why the universe looks too smooth and too flat to be an accident.

Run the plain Big Bang backwards and you find a universe that started out smooth across regions that had never been in touch, and balanced on a knife-edge between collapse and emptiness. It works, but only if the start was tuned to absurd precision. These puzzles set the stage for inflation.

1. A sky that is too smooth: the horizon problem

The cosmic microwave background has the same temperature, 2.7255 K, in every direction, to about 1 part in 100,000 (Lesson 7.4). For a cup of tea, equal temperature is no surprise: hot and cold parts mix until they agree. Mixing needs contact, though, and contact cannot travel faster than light.

When the CMB was released, 380,000 years after the start, light had only had 380,000 years to travel. Space has stretched a lot since, so that patch of “everything a point could have heard from” now looks small on our sky: about 1° across, roughly twice the width of the full Moon. Our telescopes see CMB from every direction, including spots far more than a degree apart.

A map where light always moves at 45° (distance across on a map that stretches with space, time up). Our past light cone, the path of all light reaching us today, meets the CMB at two opposite points, A and B.
Anything that ever influenced A lies in A's own past light cone. A's cone and B's cone reach the Big Bang without touching. Nothing could ever have carried heat from one to the other, yet they have the same temperature.
It is not just A and B. The whole CMB sky is made of about ten thousand patches that could never have exchanged a signal, all at the same temperature.
Preview of Lesson 11.2: if something added a long stretch of time before the hot Big Bang, the cones would reach far enough back to overlap. A and B could have grown from one region that had already evened out.

How big is one patch? On the stretching map, the distance light could travel before the CMB was released is about 280 Mpc; the CMB itself is now about 13,900 Mpc away (Planck 2018 values). So one patch spans an angle:

θH≈ηrecη0−ηrec≈28013900≈0.020 rad≈1.2∘
ηrec
the horizon when the CMB was released: how far light could have travelled since the start, measured on the stretching map. About 280 Mpc
η0 − ηrec
how far the CMB's light has travelled since, on the same map: about 13,900 Mpc
θH
the angle on our sky covered by one region that could have been in contact
Physical meaning
Two points more than about 2θH ≈ 2° apart never shared any past at all. The whole sky, 4π radians², holds about 4/θH² ≈ 104 such patches.

2. A knife-edge: the flatness problem

In Lesson 5.3 you met Ω, the density of the universe divided by the critical density. Ω = 1 means flat space. The Friedmann equation tells us how far from 1 it can drift:

Ω−1=kc2a2H2
k
the curvature of space: positive for closed, negative for open, zero for flat. It is fixed
a, H
the scale factor and the expansion rate
aH
how fast distances between galaxies grow. In a slowing universe aH falls with time
Physical meaning
While the expansion slows, aH shrinks, so |Ω − 1| grows: as a² (in proportion to time) while radiation rules, and as a (as t2/3) while matter rules. Flatness is an unstable balance, like a pencil standing on its point.

Today |Ω − 1| is below about 0.002 (Lesson 5.4). Run that backwards. At one second it must have been below about 10−18. At the Planck time, about 10−43 s, below about 10−62, and even if today’s value were as large as 1 it would still have to start below about 10−60. A start a little denser and the universe would have collapsed within moments; a little thinner and it would have flown apart before any star could form.

|Ω − 1| from the Planck time to today, for the value you choose at one second. Make it ten times bigger and today's universe is ten times more curved; a thousand times bigger and curvature takes over long before today.

3. Missing monopoles

The third puzzle comes from particle physics. Grand unified theories, which join the strong, weak and electromagnetic forces at about 1016 GeV, predict that as the universe cooled below that energy, very heavy magnetic monopoles (single north or south magnetic poles) were made: roughly one per region the size of the horizon at the time. In 1979 John Preskill worked out how many: enough to outweigh all the atoms in the universe many billions of times over. Searches have never found a single one.

This puzzle is weaker than the other two, because grand unification is itself unproven. If those theories are wrong, there is no monopole problem. But in 1979 most particle physicists expected them to be right.

4. Puzzles, not contradictions

None of this breaks the Big Bang theory. The equations work fine if you simply start them with a smooth, flat universe. The question is why the start looked like that. A theory that needs its starting numbers set to 60 decimal places explains nothing about them. Physicists prefer a mechanism that makes a smooth, flat universe the natural outcome, whatever the start. That is what inflation, the next lesson, offers.

Naming the puzzles.

  • 1956 Wolfgang Rindler defines cosmic horizons precisely.
  • 1969 Charles Misner states the horizon problem and tries to solve it with a chaotically wobbling "mixmaster" universe. It does not work.
  • 1969 Robert Dicke points out in lectures that the expansion must have been balanced extraordinarily finely early on.
  • 1979 Dicke and Jim Peebles name the flatness problem; John Preskill calculates the monopole problem.
  • Dec 1979 Alan Guth, working on monopoles with Henry Tye, notices that a burst of exponential expansion would solve all three. He publishes in 1981.

“The universe was tiny at the start, so everything was in contact.” Smaller, yes, but also younger. In a slowing universe the horizon (how far light has gone) grows in proportion to time, while distances between places grow more slowly, as t1/2 or t2/3. Going back in time, the horizon shrinks faster than the distances, so earlier on, the parts of space we now see were even less in touch.

  • Horizon problem: when the CMB was released, one causally connected patch spans only about 1° of our sky, so the CMB is made of about 104 patches that never exchanged a signal, yet all have the same temperature to 1 part in 100,000.
  • Flatness problem: Ω − 1 = kc²/(a²H²) grows as the expansion slows (∝ t with radiation, ∝ t2/3 with matter). To be below 0.002 today, |Ω − 1| had to be below about 10−18 at one second and about 10−60 at the Planck time.
  • Monopole problem: grand unified theories predict heavy magnetic monopoles that would outweigh everything else; none has been found.
  • These are puzzles about starting conditions, not contradictions. Inflation (Lesson 11.2) was invented to explain them.

Check yourself

1What is the horizon problem?
2Why is a nearly flat universe today a puzzle?
3Why is the monopole problem only a problem if grand unified theories are right?

Sources

Your ticks stay in this browser. Save them as a file on the course home.

Updated 2026-11-22. Numbers are rounded for teaching; measured values come from the sources listed in each lesson. Simulations run in your browser; nothing you type is sent anywhere.