---
title: "The small temperature differences in COBE’s sky"
description: "How did a nearly uniform microwave sky reveal the beginnings of cosmic structure?"
canonical_url: "https://sandboxphysics.com/stories/astronomy-gravity/cobe-sky-anisotropy/"
language: "en"
page_type: "story"
alternate_languages:
  zh-CN: "https://sandboxphysics.com/zh/stories/astronomy-gravity/cobe-sky-anisotropy/"
last_modified: "2026-10-24"
---

[Measuring the history of expansion](https://sandboxphysics.com/stories/astronomy-gravity/#cosmology)

Astronomy · Cosmology · Gravity

# The small temperature differences in COBE’s sky

Sandbox Physics · 2026-10-24

How did a nearly uniform microwave sky reveal the beginnings of cosmic structure?

1. Compare two sky directions
2. Build maps from repeated differences
3. Remove motion and foreground signals
4. Measure the remaining angular structure

The 1992 discovery concerned large angular scales. Fine-scale acoustic structure was measured by later experiments.

## A perfectly smooth background left a difficult question

In April 1992, George Smoot and the COBE team announced that the microwave sky contained tiny temperature differences. On top of a background near 2.7 kelvin, some broad regions were warmer or cooler by only a few tens of millionths of a degree. These differences gave cosmologists a long-sought connection between an almost uniform early universe and the richly structured one around us. The 2006 Nobel Prize later recognized this result together with COBE’s measurement of the background spectrum.

Sources: [1] [5]

## The instrument compared two pieces of sky at once

COBE’s Differential Microwave Radiometers, usually shortened to DMR, measured differences between directions. Each receiver compared a pair of horn antennas pointing sixty degrees apart. As the satellite spun and moved along its orbit, the pair sampled many combinations of sky regions. Repeated overlapping comparisons let the team reconstruct a map of temperature differences across the sphere.

This arrangement aimed directly at the small quantity of interest. The receiver did not have to remeasure the full background temperature with extraordinary accuracy at every point. It asked which of two broad patches was slightly brighter. DMR observed at 31.5, 53 and 90 gigahertz, with two nearly independent receiver channels at each frequency. The channels provided ways to compare results and to investigate noise or signals that changed differently with frequency.

Sources: [2] [3]

## Our motion paints a much larger pattern

The first microwave map contains more than primordial temperature variations. The Solar System moves relative to the background. Looking into that motion, the radiation appears a little warmer; looking backward, it appears cooler. This broad hot-side/cold-side pattern is called the dipole. Its scale is a few thousandths of a kelvin, much larger than the small variations the team sought.

Radiation from the Milky Way adds another pattern, concentrated near its bright band across the sky. The team compared frequencies and excluded strongly contaminated regions in key analyses. After accounting for the dipole and Galactic emission, much smaller patches remained. The colour range of the familiar map is greatly stretched to reveal them. Red and blue encode slight temperature differences; the image’s oval shape is a projection of the whole sky onto a flat page.

Sources: [1] [2]

## Independent receivers tested whether the patches belonged to the sky

Instrument noise can draw apparent mottling on a map. A useful check is whether independent receivers show corresponding structure in the same directions. Adding two channel maps retains their shared sky signal along with noise. Subtracting them suppresses the common sky and reveals the differences associated with measurement noise and other channel-specific effects. Comparisons across observing periods provide another check.

The first-year paper reported an overall root-mean-square variation of about 30 microkelvin, with an uncertainty of 5 microkelvin, after smoothing to ten-degree resolution, removing the dipole and selecting high Galactic latitudes. Root-mean-square variation is one measure of a pattern’s typical amplitude. The smoothing and sky selection are part of that number. A differently smoothed map naturally gives a different amplitude because averaging neighbouring directions erases some structure.

Sources: [1]

## The angular scale carries information too

DMR’s beam was about seven degrees wide, so it saw broad patches. A wider beam averages more sky into each reading and blurs out smaller patches first. This is why an angular power spectrum, which groups a map’s variation by patch size, is useful. It describes how much structure survives on large or small angular scales. The map gives locations; the spectrum summarizes their scale-dependent strength.

Cosmologists connected these weak variations with early differences in matter density and the gravitational fields through which the photons travelled. Slightly denser regions can attract more material and grow as the universe evolves. The microwave temperature is not a direct photograph of mass density, so a physical evolution model is needed for the connection. COBE supplied large-scale amplitudes that such models had to explain. Later experiments resolved much finer structure and tested additional details of the early plasma.

Sources: [1] [4] [2]

## The texture made the almost-uniform universe measurable

The discovery connected an almost uniform early sky with the growth of cosmic structure. The background was extremely uniform, yet galaxies required departures from uniformity to grow. COBE detected small departures on large scales and measured their size. The same faint radiation that supported a hot early history now also supplied a starting point for calculating how cosmic structure could develop.

Sources: [1] [4]

## Explore the related models

Choose a model to explore the physical idea.

### [Blur a synthetic microwave sky](https://sandboxphysics.com/labs/cmb-sky-angular-spectrum/)

Widen the receiver beam and watch smaller patches fade.

The synthetic degree-two-through-twelve sky and its power spectrum share the same harmonic coefficients.

## Sources and reading notes

1. [Smoot et al. (1992), Structure in the COBE Differential Microwave Radiometer First-Year Maps](https://cosmos.lbl.gov/PDFs/1992ApJ___396L___1S.pdf)

   The first-year anisotropy detection; read the smoothing, dipole removal, Galactic cut and 30 ± 5 microkelvin RMS together.

2. [NASA LAMBDA, The COBE Differential Microwave Radiometers](https://lambda.gsfc.nasa.gov/product/cobe/about_dmr.html)

   Six receivers, three frequencies, sixty-degree differential pointing and seven-degree beams.

3. [NASA LAMBDA, COBE DMR Sky Maps and Analyzed Science Data Sets](https://lambda.gsfc.nasa.gov/product/cobe/dmr_products.html)

   Map reconstruction from differential observations and the independent frequency/channel map products.

4. [Peebles (2019), Nobel lecture: How Physical Cosmology Grew](https://www.nobelprize.org/uploads/2024/02/peebles-lecture-2.pdf)

   The author connects recombination, microwave fluctuations, galaxy clustering and the history of dark-matter cosmology.

5. [The Nobel Prize in Physics 2006](https://www.nobelprize.org/prizes/physics/2006/summary/)

   Mather and Smoot: blackbody form and anisotropy, corresponding to complementary COBE measurements.

[How we learned to read the universe](https://sandboxphysics.com/stories/astronomy-gravity/) [COBE compared the sky with a laboratory blackbody](https://sandboxphysics.com/stories/astronomy-gravity/cobe-blackbody-spectrum/) [From an almost smooth sky to a universe full of galaxies](https://sandboxphysics.com/stories/astronomy-gravity/peebles-growing-universe/) [The microwave hiss that remained after the antenna was checked](https://sandboxphysics.com/stories/astronomy-gravity/penzias-wilson-microwave-background/) [Distant supernovae showed that cosmic expansion had accelerated](https://sandboxphysics.com/stories/astronomy-gravity/supernova-accelerating-universe/)