The very first exoplanets were finally discovered in the early 1990s, and since then over 4,000 have been caught floating about in the inky pool of deep space. It’s clear that the cosmos is swimming with alien worlds, and some are pretty damn peculiar.

In recent years, astronomers have confirmed the existence of so-called “cotton candy” planets: portly, Jupiter-size worlds that are remarkably lightweight. These oddities have densities comparable to cotton candy, hence the moniker. Saturn is famous in our own solar system for being of such low density that it could float in a big enough bathtub, and these exoplanets make our stellar neighborhood’s gas giant look like a lead weight in comparison.

Scientists were already fairly bemused by these strange worlds. And then came along WASP-107b, the cotton candy world that shouldn’t exist.

Discovered in 2017, this planet was seen orbiting a star (WASP-107) 212 light-years from Earth. Being the same size as Jupiter but 10 times lighter, WASP-107b turned out to be another cotton candy world. But the plot thickened when, as reported in a recent study in The Astronomical Journal, researchers determined its core was no more massive than four Earths – nowhere near enough, it seemed, to use its own gravity to gather all that puffy gas it has in its atmosphere.

It was also clocked orbiting its star at a distance 16 times closer than Earth is to the Sun. That made no sense, because if it originally appeared in that spot its atmospheric envelope would have been obliterated. “It could never have formed there,” says Caroline Piaulet, a doctoral student at the University of Montreal’s Institute for Research on Exoplanets and the new study’s lead author. Naturally, everyone wanted to know the answer to one question: what the hell, WASP-107b?

Some astronomers don’t seem overly keen on cotton candy worlds. They don’t fit in with what they know to be true about planetary systems. One abstract from 2019 actually says “they shouldn’t exist and yet we’ve already detected half a dozen of them with Kepler alone,” something I can only imagine as being said with a frown.

“This is a recurring trend in exoplanet science,” says Nikku Madhusudhan, an exoplanetary science expert at the University of Cambridge. Scientists can be stuck in their ways based on apparently ironclad rules born from a limited data set. They are “thumping on the table, saying this is how it ought to be – and then next year we discover an object which completely defies what they believe.”

Cotton candy worlds are one such type of object, with their extremely puffy hydrogen and helium atmospheres lacking satisfactory explanation. WASP-107b is only going to muddy the waters further.

Piaulet and her colleagues were fascinated by this ludicrously lightweight world and wanted to dive deeper. They used the Keck Observatory in Hawaii to more precisely determine its mass, tracking the wobble of its star to determine just how much planet was pulling on the furnace as it revolved around. Then, they simulated the possible evolutionary pathways the planet may have taken to get to the mass and size it was today over the course of three billion years, the age of its star.

For giant planets, it’s been assumed – partly based on old theories of how the solar system’s gas and ice giants came about – that it takes 10 Earth-masses of solid material to scoop up gas fast enough to build Jupiter-size and Jupiter-mass planets. But WASP-107b’s internal structure didn’t seem to add up. “It has such a low mass that its core couldn’t be more massive than about four times the mass of the Earth, which is way lower than what we thought was necessary to be able to accumulate that much gas in the first place,” says Piaulet.

So how did this exoplanet’s tiny core bulk up? The most parsimonious idea is based on a key concept: cooler planets trap gas faster.

As gassy planets form around a baby star, they do so around a solid core of fairly rocky stuff. That core has a gravitational influence, and any gas that falls within a certain radius gets gobbled up and becomes atmosphere.

The somewhat violent way planets form means they start off quite hot. And the gas they accrete, says Piaulet, acts like an insulating blanket, one that prevents them from cooling quickly. And hotter planets can’t gather and condense gas very well, which stops them from getting particularly puffy.

WASP-107b, then, could never have formed right in its star’s face. With such a teeny core, and in the midst of such high temperatures, there was no way it was able to steal and retain so much gas.

A clue to its secret was found in the world’s unusual dance. Instead of orbiting in a flat plane, like marbles all sitting on the same table, its path takes it close to the star’s poles. Stranger still, it’s orbiting the star backwards, in that it’s going in the opposite direction to the star’s own rotation.

“In the solar system, all of the planets orbit in the same direction as the Sun is rotating,” says Ryan Rubenzahl, an astrophysics doctoral student at Caltech and study co-author. “If something is orbiting retrograde, it’s an indication that it didn’t form like that.” Every planet should come from the same primordial disk of building materials, so for it to be going the wrong way suggests something strange happened to put it there.

The only way this made any sense was if WASP-107b was born on the fringes of its planetary system, far from its star. As prior work has suggested, if you build a planet out there, the gas, already frigid, can contract around the young world faster, giving the planet a chance to gather even more.

Another factor may have been at work, says Piaulet. This gas is often dusty, which makes a planet’s atmosphere somewhat opaque – excellent for trapping heat. WASP-107b probably had the dust washed out of its skies early on, allowing heat to radiate into space, cooling the planet quicker and letting it snatch up more gas.

“If you have a low-opacity gas and you place the planet further away from the star, it’s gonna be able to accrete [gas] faster,” says Piaulet. “And that’s the most likely explanation we found for how this planet formed.” If this is true, then it may help explain why there are so many gas giant exoplanets out there: there are simply several different ways of making them, some weirder than others.

A companion world, WASP-107c, hints at how the super puffy world ended up spiraling so close to its star, and ending up in the precarious position it’s in today. The personal space-invading WASP-107b orbits its star once every 5.7 days, but this second, much heavier and considerably more distant planet orbits in a notably elliptical pattern once every three years.

Over eons of time, the orbits of planets close to their stars should become increasingly circular as their gravitational tango with their star smooths things out. Being far away, WASP-107c has preserved its eccentric orbit. Consequently, it has retained the memory of the chaos that, says Piaulet, “brought its little sister inside” close to the planetary system’s heart. It’s not clear what happened, but perhaps a close encounter between the two worlds, or with another planet that was summarily ejected from the system, led to the weird arrangement we see today.

This story is an elegant first shot at explaining the system, says Madhusudhan – a “great place to start.” But the details of the saga could change the more we learn about other exoplanetary systems.

That, indeed, is the problem with studying exoplanets for a living. The cosmos is full of puzzling sights, including super dense worlds lacking any atmosphere whatsoever, rogue planets without a star zipping through the galaxy and, yes, cotton candy worlds. Even after a very brief survey of the universe, says Rubenzahl, it’s clear the solar system-centric model of planet formation doesn’t work very well.

It took centuries to come up with a working model for the whole solar system. “And now we have thousands of solar systems, and they don’t look like ours,” says Madhusudhan.

Finding more exoplanets will, eventually, help us begin to answer some of our questions, including the origins of cotton candy worlds like WASP-107b. But the gold rush of discoveries is becoming a little bit much for astronomers. “The rate at which we are finding new things like this is much faster than the rate at which we’ll be able to [handle] them,” says Madhusudhan, laughing. “It’s a very human problem.”