Last month when scientists announced new evidence that in a distant galaxy, two giant black holes are spiraling toward each other, destined for a cataclysmic merger, they got a lot of attention. More than one non-journalist, non-scientist friend excitedly mentioned the news to me. (At least one of those friends missed the part about the black holes being 3.5 billion light years away, as opposed to in our backyard.)

It was a very cool finding. It also aligned in a serendipitous way with reading I’d been doing on the history of research into radio galaxies, quasars, and black holes. It was a reminder that sometimes ideas get shelved for years, even decades, before reemerging in new forms.

Here’s what I mean. The new research came from scientists at Columbia University, who analyzed data from space telescopes, including Hubble, and concluded that in the heart of the quasar PG 1302-102, two supermassive black holes will merge some 100,000 years from now. (A team from Caltech first found the pair of black holes and published their results early this year.) Quasars, some of the brightest and most distant objects ever observed, are a type of active galactic nuclei (AGN). In the 1950s, when astronomers discovered the first active galactic nuclei—radio galaxies, which are generally closer to us and less extreme than quasars—they thought they were seeing two galaxies collide. It was a good guess. Through optical telescopes, the first known radio galaxy, Cygnus A, looked like two galaxies colliding. But the picture quickly became more complicated. In 1953, astronomers showed that Cygnus A’s radio emission came from two symmetric lobes arranged in the shape of a dumbbell. Many researchers surmised, correctly, that some tiny object at the center of this dumbbell must power both lobes.

By the late 1950s, most scientists had set the galactic-collision hypothesis aside. Their new mission was to figure out what was hiding at the center of these weird objects. Candidates included “superstars” with the mass of millions or billions of suns, white holes pouring energy into our universe from some other dimension, clusters of neutron stars, and antimatter-matter annihilations. Eventually, scientists settled on the explanation that is widely accepted today: quasars, radio galaxies, and other active galactic nuclei are all powered by raging supermassive black holes.

If the Columbia and Caltech scientists are right, of course, PG 1302-102 contains not one but two central supermassive black holes. Here’s where the old idea reemerges: the reason is because PG 1302-102 is the probably the product of two galaxies merging.

Theorists revived the colliding-galaxy hypothesis in the 1980s to explain how quasars differ from other galaxies. By then, it seemed likely that most galaxies contained supermassive black holes at their cores. But only about one percent of galaxies have active nuclei. What’s different about them? What causes them to light up? Colliding galaxies, naturally.

The idea wasn’t that stars were smashing into each other and exploding: stars are so far apart that the chance of a direct collision, even in merging galaxies, is small. Instead, the theory went, when galaxies merge it causes drag and torque and general orbital disruption that knocks stars and clouds out of their usual paths, in some cases sending them spiraling down to the black hole(s) at the center of the quasar. The black-hole feeding rampage that follows generates the geysers of radiation that we here on Earth identify as quasars.

Zoltan Haiman, one of the Columbia researchers, said that the colliding-galaxy model of quasars has held up fairly well since the 1980s. “There has been a lot of theoretical work since the 1980's, particularly more sophisticated simulations, which show that when two galaxies merge, a lot of gas is driven to the nucleus, and that this can light up a quasar,” he wrote me in an email. “There has also been a lot of observational work, which has shown that quasars are associated with bursts of star-formation (another sign of a recent merger) and also the hosts galaxies of many quasars (these hosts are generally very hard to see) look like galaxies in the process of merging, or a single galaxy that shows a disturbed shape, indicative of a recent merger.” The picture these days is more nuanced, however, Haiman said. The biggest, brightest quasars are probably the result of galactic collisions, but for smaller galaxies and dimmer quasars the situation is less clear. 

As tends to be the case with phenomena at the edge of the observable universe, questions remain. Among them: How many quasars contain multiple supermassive black holes? Do those black holes eventually merge, or do they keep orbiting each other indefinitely?

PG 1302-102 suggests answers. “We are almost 100 percent sure that most galaxies have such a pair [of black holes] at some point,” Haiman told me over the phone. “This follows directly from the fact that we know that each galaxy usually has a black hole at the center—a 20-year-old discovery. A: all galaxies have black holes at the center. B: galaxies merge all the time. From these two things it’s completely obvious that often there would be black hole mergers. Each galaxy you look at most likely merged with another galaxy in the past few billion years. The question is, have the two black holes merged, or are they still there?”

The reason to think that orbiting pairs of black holes might never merge is called the “final parsec problem.” The idea is that if you have a pair of orbiting black holes that are far apart, they have no reason to merge unless something degrades their orbits, bringing them gradually closer. You need friction to make that happen. But what if there’s nothing around to create that friction? How do these black holes ever get close enough to merge? This has been a surprisingly vexing problem. (So has the question of how matter orbiting a black hole gets down to the event horizon where it can be eaten. That’s a subject for another post.) “Some people think the final parsec problem cannot be solved—that the black holes get hung up [orbiting and can never merge],” Haiman says. “That is another important thing about our paper. These black holes are separated by one percent of a parsec. Clearly they managed to solve the infamous final parsec problem.”