If you could see the Milky Way edge-on, it would not look like one flat sheet of stars. It would look like two disks stacked together. The thin disk holds the young and Sun-like stars and the spiral arms. Around it sits a puffier thick disk made of ancient, metal-poor stars. Astronomers have argued for a long time about how this layered structure came about. A new paper in The Astrophysical Journal Letters offers one of the clearest answers so far, and it rests on something astronomy has always struggled to measure: how old a star actually is.

The study is led by David Mirabal, a pre-doctoral researcher at the Instituto de Astrofisica de Canarias (IAC). It concludes that the disk settled in two distinct stages. First it flattened very rapidly. Then, from roughly 10.85 billion years ago, it thinned slowly. That timeline supports a picture in which the galaxy built itself "upside-down," in which the thick disk formed early, rather than one dominated by later heating.

The two competing stories

There have been two main ways to explain a galaxy with a thick disk wrapped around a thin one, according to Universe Today's coverage of the work.

The first is dynamical heating. In this version, stars are born in a thin disk and get puffed up over time. Gravitational disturbances gradually push them onto more inclined orbits, so old stars end up in a thicker layer simply because they have been knocked about for longer.

The second is upside-down formation. Here the early galaxy was turbulent, and the stars that formed then were born in a thick, puffy configuration. As conditions calmed, star formation moved into a progressively thinner layer. The thick disk is then a fossil of the galaxy's youth, not the product of later heating.

The paper frames the problem as separating these two scenarios, which has been hard without precise ages for large, representative samples of stars. What helps separate them is the shape of the relationship between age and thickness over time: how fast the thickness changes, and whether there is a clear break. Measuring that shape requires stellar ages good enough to split the disk into narrow age slices. This is the hard part.

Dating stars to within 5 percent

The team used data from the third data release of the European Space Agency's Gaia mission (Gaia DR3, published in June 2022). They did not try to age-date stars one at a time. Instead they applied a technique called colour-magnitude diagram fitting. A colour-magnitude diagram plots stars by brightness and colour. A population that formed at a particular time with a particular chemical makeup traces a characteristic pattern on that plot. By fitting the observed diagram for complete samples of stars in the solar neighbourhood, the researchers recovered age-metallicity distributions. In effect, these are maps of how many stars formed when, and with what composition.

From those distributions they built what the paper calls "quasi-mono-age-metallicity populations": groups of stars that share nearly the same age and metal content. According to the paper's abstract, the age resolution is better than 5 percent. That works out to about 0.5 billion years even for the oldest stars, where age estimates are usually at their least reliable. The method is part of a project the authors call ChronoGal.

With populations sorted that finely, the team could measure the scale height of each group. Scale height is a standard measure of how far a population of stars extends above and below the galactic midplane. They then tracked how scale height changes with age.

A fast collapse, then a slow squeeze

The first result is a simple one: scale height correlates with age. Older populations sit in thicker layers. The more revealing result is how that thickness changed over time.

The abstract reports that scale height dropped from more than 1.26 kiloparsecs to about 0.5 kiloparsecs "in merely 2 Gyr." Universe Today gives those figures as roughly 4,100 light-years shrinking to about 1,600 light-years. After that rapid phase the thinning becomes gradual, with the youngest populations sitting at a scale height of about 0.1 kiloparsecs.

The abstract puts the break between the two regimes at 10.85 billion years ago, with an uncertainty of about +0.13 and -0.12 billion years. The same text is in the arXiv preprint, which was submitted on October 2, 2026.

The pattern is two different rates of change with a clear break between them, and that is what makes it informative. The authors read a rapid early contraction followed by a long, slow tail as the signature of a system that started out chaotic and then calmed down.

What the authors conclude

The paper says the extremely rapid early settling points to a "fundamentally upside-down formation process at early times," when the thick disk was basically formed. In the authors' reading, the disk went through two phases. An initial, highly turbulent epoch gave birth to the thick disk, and a more quiescent period of gradual gas settling followed in which the thin disk formed. They call this the "two-stage settling" in the paper's title.

The authors describe the later phase as one of secular evolution and gradual gas settling. Their claim is that the thick disk was basically formed early, by the conditions stars were born in.

Why It Matters

The thick disk/thin disk question is not only about the Milky Way's appearance. It is a test of how disk galaxies assemble in general. If the thick disk records a turbulent early phase, then the stars in it preserve information about what our galaxy was like in its first few billion years. That includes how quickly it settled and how violent its youth was.

The work also shows what precise ages can do. The paper notes that the lack of precise ages for large, representative samples has traditionally been a drawback in telling formation scenarios apart. Ages good to better than 5 percent, drawn from complete local samples, let the researchers see a break in the disk's history. If the approach holds up, it may help astronomers turn Gaia data into a timeline rather than a snapshot.

There are limits. The samples come from the solar neighbourhood, so whether the same two-stage history holds across the whole disk is a question for future work. As always with a new result, independent analyses using different age-dating methods will be the real test.

Sources