As starlight travels through the air, it leaves its mark: water vapor absorbs some of the light, creating dark bands in the spectrum, like gaps in a text. For six years, a telescope in the Chilean desert recorded starlight for entirely different purposes, but those recordings were preserved. By reading the dark bands, you can determine how much water hung above the observatory on each clear night. And that's exactly what they did: they took 5879 spectra from 918 nights and, from each band, calculated the thickness of the water layer that would result if all the vapor were condensed.

For the future European Extremely Large Telescope, water in the air is like frosted glass: it steals infrared light and hinders the view of distant galaxies. So it's crucial to know how dry the sky is above Cerro Armazones. It turns out that on average, 2.43 mm of precipitable water hangs there, while on the neighboring mountain Cerro Paranal, it's 2.96 mm. The 22 percent difference is larger than can be explained by the difference in altitude alone, meaning the site is drier due to local winds. Moreover, the water is evenly distributed across the sky, with no imbalance between east and west.

This measurement method requires no new instruments: the telescope is already there and recording light; you just need to learn to read its data. This is important not only for the giant telescope but also for the future Cherenkov Telescope Array, which will catch particles from space and needs precise humidity corrections. Now it's clear that global weather models are inadequate for this corner of the Atacama, and either local refinements or direct measurements from starlight are needed. The old recordings have become a chronicle of the sky, and it already shows that the Armazones site is indeed dry, so the giant eye will have something to breathe.