Astronomers have now found more than a thousand stars with more than one planet. Taken as a group, those systems look like a mess: tight clusters of small worlds closer to their star than Mercury is to the Sun, sparse arrangements of giants on wide orbits, and plenty in between. A new study in Nature Astronomy, published Oct. 13, 2026, argues that the variety hides a simple split. Measured the right way, multi-planet systems fall into two distinct families, and which family a system joins appears to depend mostly on its star.

The team was led by Italy's National Institute for Astrophysics (INAF), with Diego Turrini (INAF) and Camilla Danielski (University of Valencia) as co-first authors. Co-authors include Danae Polychroni, Diederik Kruijssen, Zinzi, Goodwin, Longmore, Chevance and Ramambason, among others. They report a clear bimodality in total system mass. On one side are compact, lower-mass systems around stars up to roughly the Sun's mass. On the other are heavy systems around more massive, metal-rich stars, and these are about an order of magnitude more dynamically agitated.

How the sorting was done

Planet surveys don't sample the galaxy evenly, and that is the main obstacle to any study like this. Some kinds of planets are much easier to detect than others, so a pattern in the catalogue can reflect the telescopes rather than the planets. According to INAF's announcement, Danielski said the main difficulty in the analysis was determining whether the two families were real or an illusion created by the limits of current observing techniques. The paper's abstract describes the defence the team chose: "architecture-agnostic" metrics, meaning measures that don't assume in advance what a planetary system should look like. These were applied to four independent samples of systems.

The team did not use every known system. It worked with the roughly 300 best-characterised ones out of the 1,000-plus multi-planet systems discovered so far, per INAF and Space.com. That gives up sample size in exchange for reliable masses and orbits, which is the right trade when the result depends on adding up how much planet each system holds.

When the system masses were plotted, they formed two humps instead of one smooth spread. That bimodality is the core result, and the paper reports it across all four samples.

Family one: the quiet, tightly packed systems

The lower-mass family orbits mostly subsolar-to-solar stars, meaning stars from somewhat smaller than the Sun up to about its size. Their planets are tightly packed and fit the pattern astronomers call "peas in a pod." EarthSky sums these systems up as having less massive planets, but more of them, with a less violent history. INAF says the planets mostly orbit much closer to their stars than Mercury does to the Sun.

The study measures how agitated each system's orbits are, a quantity called dynamical excitation. For this family, the most common value is comparable to the Solar System's. In that one respect, these crowded inner systems are about as calm as ours.

Family two: heavy, hot and probably unstable from the start

The second family looks very different. According to the paper, these systems form in metal-rich environments around solar-to-supersolar stars. INAF says their total planetary mass is often 2 to 10 times that of the Solar System. EarthSky describes them as having fewer, more massive planets on wider and more varied orbits, with collisions and ejections possible along the way.

The paper calls this family "gravitationally compact" and an order of magnitude dynamically hotter than both the Solar System and the lower-mass family. The authors read that heat as evidence of global instabilities and planet-planet scattering, where gravitational encounters between big planets reshape orbits and can throw worlds out of the system altogether.

The team also has a view on when that happens. As reported by INAF, Polychroni says the scenarios that best reproduce the wide, massive systems are those in which the systems are unstable from birth, because the greater abundance of planet-forming material makes the planets form too large for the system to stay stable. Turrini adds that the truth lies somewhere in the middle between traits set at birth and those sculpted by evolution over time.

A "genetic link"

Turrini describes the result as a "genetic link" between planetary systems and their stars, a framing reported by both INAF and Space.com. Two stellar properties line up with the split. One is mass: the heavy family orbits stars from about solar mass upward, and the compact family orbits stars up to about solar mass. The other is metallicity, the share of elements heavier than hydrogen and helium. The heavy family forms in metal-rich environments.

The paper describes a link between stellar properties and system architecture, and says the architectures arise from the combined influence of formation processes and host star properties. What that suggests is that once you know a star's properties, you can say a lot about what kind of planetary system it probably hosts and how turbulent that system's past was likely to be.

Why It Matters

A clean two-way split in system mass, seen across four independent samples, gives theorists a concrete target to test. Formation models now need to produce two populations: one that builds many small planets close in and leaves them mostly undisturbed, and one that builds a few heavy planets and then rearranges them violently, with instability that may begin at birth.

The result also matters for how future observations are planned. INAF and Space.com both say the finding could help point astronomers toward systems with potentially habitable planets. If a star's mass and composition predict whether its planets lived through a calm history or a period of scattering and ejection, surveys could weigh that before any observing time is committed.

One caution: the result rests on the best-characterised 300 systems, not on all known systems. The authors chose metrics and multiple samples specifically to limit observational bias, and Danielski addressed selection effects directly. Still, the boundaries of the two families will be tested as more systems get measured well enough to join the sample.

The work was supported by an International Space Science Institute (Bern) project led by Diederik Kruijssen. INAF lists the University of Valencia, the Italian Space Agency (ASI) and Uppsala among the partner institutions.

Sources