In the early 1960s two rival accounts of the universe's origin stood opposed: a steady-state model, expanding through continuous creation and always looking the same, and a Big Bang model implying a hot dense beginning. The latter predicted that radiation left from that beginning should reach us today in the microwave band, arriving equally from every direction. Nobody was looking for it.

Arno Penzias and Robert Wilson were calibrating a horn antenna for radio astronomy at Holmdel, New Jersey. Whichever way they pointed it, a steady hiss refused to go away. For a year they eliminated possible sources — electronics, atmosphere, city noise, even the pigeons nesting inside and the droppings they left. The noise remained. Thirty kilometres away at Princeton, Robert Dicke's group was building apparatus to search for exactly such radiation; the two groups made contact, and in 1965 two papers appeared side by side, one reporting the measurement, the other its meaning.

The measurement effectively ended the steady-state model. Later satellites — COBE (1989), WMAP (2001), Planck (2009) — mapped this radiation down to fluctuations of one part in a hundred thousand, and those fluctuations are the seeds of today's web of galaxies. The age of the universe, its matter-energy composition and its geometry are largely computed from these maps. That a piece of engineering noise turned cosmology into an observational science is told nearly as often as the discovery itself.