The Milky Way was not always flat. Its youngest stars stay within about 400 light-years of the midplane, while its oldest stars range through a layer roughly ten times thicker. How the Milky Way disc went from a puffed-up swarm to a thin, orderly sheet has been argued over for decades. A new study of stellar ages built on Gaia data suggests the story has two chapters: a fast, turbulent thinning that ended about 10.8 billion years ago, then a slow settling that has not stopped since.

David Mirabal, a doctoral researcher at the Instituto de Astrofísica de Canarias (IAC) and the University of La Laguna, led the work, which is published in The Astrophysical Journal Letters. The IAC announced the result on 19 October. It comes from ChronoGal, a European Research Council project that aims to date the Galaxy’s stars well enough to read its history as a timeline. The raw material is the European Space Agency’s Gaia mission, which has measured the positions, motions and brightness of more than a billion stars.

Ages are the hard part. A star’s position and speed can be measured directly; its age has to be inferred. The team fitted colour-magnitude diagrams, plots of brightness against colour that sort stars by mass and stage of life, to complete samples of stars around the Sun, up to 1 kiloparsec (about 3,300 light-years) above and below the plane. Matching those diagrams against models of stellar evolution gave ages with uncertainties below 5%, around 500 million years even for the oldest stars. That is sharp enough to separate generations of stars rather than lump them into broad old and young bins.

With ages in hand, the researchers grouped stars of similar age and metal content and measured how far each group spreads above and below the disc, a quantity called the scale height. The trend was clear but not smooth. Among the oldest groups, scale height falls steeply with time, from more than 1.26 kiloparsecs (about 4,100 light-years) to roughly 0.5 kiloparsecs (1,600 light-years) over about two billion years. Then the slope changes abruptly, at a turning point the fit places 10.85 billion years ago. After that, each new generation is only slightly thinner than the one before, down to about 0.1 kiloparsecs for stars forming today.

That elbow matters because astronomers have had two competing explanations for why old stars ride so high. In one, the disc was born thin and its stars were later kicked upwards by encounters with gas clouds, spiral arms and passing satellite galaxies, a slow process called dynamical heating. In the other, known as upside-down formation, the early gas was turbulent, so the first stars formed in a thick layer, and each later generation formed in a thinner one as the gas calmed. Simulations can produce either picture. Observations had not been able to choose.

Here the steep early trend tips the balance. Heating alone struggles to explain how quickly the oldest populations thinned, so the authors read the first phase as gas settling down, with the thick disc forming in the turbulent period and the thin disc in the calmer one that followed. Mirabal describes this in the release as the first clear observational evidence favouring one model over the other. For stars younger than the turning point, the data fit either explanation.

Its timing invites speculation. The break falls close to the Gaia-Sausage-Enceladus merger, when a smaller galaxy fell into the young Milky Way roughly 10 billion years ago. The paper notes the coincidence but does not claim cause and effect. Repeated early mergers, the way fresh gas arrived, and energy from intense bursts of star formation are all candidates for keeping the young disc stirred up.

There are real limits. The sample covers only the solar neighbourhood, so the inner and outer disc may have settled on different schedules. The slope for the oldest stars is loosely constrained, and the scale heights of the two oldest groups could only be estimated roughly. Even the paper quotes the length of the fast phase as anything from two to three billion years in different sections. An earlier study using different methods found smooth age trends with no clear break. Grouping stars on a simple grid instead of by clusters softens the early slope here too, although the turning point survives.

If the result holds, the Sun is a child of the quiet chapter, born into a disc that had long since stopped thrashing about. The old stars that drift thousands of light-years above it are not strays knocked off course. They are fossils of a younger, rowdier Galaxy, still riding at the heights where they were born.