At the South Pole, where the air is dry and transparent to millimeter waves, stands a telescope that looks not at stars but at the oldest light in the universe. This light is the cosmic microwave background — the radiation left over from the Big Bang, filling the entire sky. Like an old photograph, it captures tiny temperature irregularities — imprints of quantum fluctuations stretched to cosmic scales by inflation.

The SPT-3G telescope collected this light for two years at three frequencies—95, 150, and 220 GHz — covering a sky area of 1,646 square degrees, roughly equivalent to eight full moons. The scientists measured how much the background temperature varies from point to point on angular scales from 2 to 13 arcminutes. These variations, or the power spectrum, carry information about what happened to the radiation on its way to us: hot gas in galaxy clusters slightly heats the photons (the Sunyaev–Zel'dovich effect), radio emission from galaxies adds its own noise, and dust in distant galaxies glows in the infrared.

To isolate the contribution of each source, the scientists had to build complex models. It turned out that the result strongly depends on how the thermal Sunyaev–Zel'dovich effect and the cosmic infrared background correlate with each other. When the authors allowed the angular dependencies of these signals to vary freely, they obtained at ℓ=3000 (corresponding to an angle of about 3.4 arcminutes) a thermal effect power of 4.91±0.37 μK² and a kinematic effect power of 1.75±0.86 μK² at 143 GHz. These numbers are like fingerprints that reveal how much hot gas was in clusters and how fast it was moving.

The kinematic Sunyaev–Zel'dovich effect is especially interesting: it arises from the motion of clusters as a whole and does not depend on gas temperature. By measuring its power, the scientists estimated how long the epoch of reionization lasted — the period when the first stars and galaxies ionized the neutral hydrogen that filled the universe after it cooled. It turned out that the transition from 25% to 75% ionization took less than Δz=3.8 (according to a semi-analytic model), and from 5% to 95% less than Δz=6.1 (according to AMBER simulations). This means that reionization was quite rapid: as if you lit a candle in a dark room, and it illuminated the entire space in a short time, rather than smoldering gradually.

These constraints are important for understanding when and how the first stars and galaxies formed. If reionization was rapid, it means stars were born in large numbers and efficiently ionized the surrounding gas. If it had stretched over hundreds of millions of years, the scenario would be different. The new SPT-3G data narrow the field for theorists, forcing models of the early universe to be more precise.

The telescope at the South Pole continues to accumulate data, and in the future, adding more years of observations, scientists will be able to measure the power spectrum even more accurately and perhaps see fine details that are currently hidden behind noise. For now, each new batch of data is like developing an old photograph: the image becomes clearer, and we see what the universe looked like in its first moments.