Astronomer Bob Benjamin has spent the past 20 years trying to figure out what the Milky Way looks like. The work isn't easy, because we're inside the galaxy and can't see it from the outside, but astronomers have ingenious workarounds, and Benjamin thinks “it's a knowable thing.” He carries in his mind a picture of what astronomers have been able to put together so far: a dense, barred center embedded in a layered disk of gas and stars, some of which pile up into arms that spiral through the disk, all encased in a sparse spherical halo of stars.
Piecing together this much of a Milky Way map has been so difficult that during interviews Benjamin and other astronomers repeatedly cite the story of the blind men and the elephant: men who cannot see each touch an elephant's trunk, ear or leg and respectively describe a snake, a fan or a tree trunk; they miss the whole elephant entirely. At least astronomers knew what they didn't know. They knew stars in different parts of the galaxy were different ages, but they couldn't account for why. They knew stars formed in gigantic clouds of gas, but the clouds were all but unmappable. They'd seen other galaxies merging with one another and looking unkempt, but they didn't know whether an earlier Milky Way might have done the same. When he began his career, Benjamin figured that the galaxy was in equilibrium, stable since birth, orderly and elegant.
But that picture has changed in recent years as scientists have begun systematically mapping stars wholesale. The bounty of data comes from a batch of new surveys, most notably one by the European Space Agency (ESA) observatory Gaia, that are collecting stupefying amounts of information. As of 1993, ESA's previous star-mapping satellite, Hipparcos, had mapped 2.5 million stars; by 2023 Gaia had mapped around 1.8 billion of them.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Gaia, which released its first data in 2016, has been complemented by an acronym soup of other telescopes and surveys—especially the Sloan Digital Sky Survey's (SDSS) Apache Point Observatory Galactic Evolution Experiment (APOGEE) and its just started Milky Way Mapper (MWM), as well as the Radial Velocity Experiment (RAVE), Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST), GALactic Archaeology with HERMES (GALAH), Atacama Large Millimeter/submillimeter Array (ALMA), and Hectochelle in the Halo at High Resolution (H3). Collectively these projects have gathered images and spectra—stars' light spread out into its individual wavelengths—for millions of stars.
With all these data, astronomers are making the first exact maps of the Milky Way: locations of stars in three dimensions, plus a record of their motions made by repeatedly imaging them over time. The result is a deep, high-resolution movie of a few billion swirling stars that helps to reveal not only the galaxy's structure but also its surprisingly tumultuous history, along with the histories of its stars and the galaxy's means of making more stars. It's “the single largest increase in astronomical knowledge in, like, forever,” says Charlie Conroy of Harvard University. “It's been shocking.”
In short, the maps show not the Milky Way in static equilibrium, as researchers expected, but rather the galaxy's departure from it. As Benjamin, an astronomer at the University of Wisconsin–Whitewater, says, “Oh, my god, it's real. And oh, my god, it's a mess.”
Of course, mapping the stars is nothing new. Around 4,000 years ago ancient Mesopotamians watched the sun, moon and wandering planets move against the stars of the zodiac's constellations—the True Shepherd of Heaven, the Old Man, Pabilsag (a scorpionlike god), the Goat-Fish, the Hired Man—which they believed were laid out by the great god Marduk so people could arrange their lives and organize the year. If a baby was born in the month when the moon was in the Bull of Heaven, and the moon then moved night by night through the rest of the constellation creatures of the zodiac until it came full circle back to the bull, then the baby would be one month old. To track these motions, Mesopotamian astronomers looked at specific stars, which later scholarship called Normal stars, and held up fingers to measure the daily distances between the Normal stars and the moon, sun and planets.
By around 120 B.C.E. Greek astronomer Hipparchus had replaced fingers held against the sky with a universal grid of longitude and latitude on which stars could be located. Beginning in the early 1600s and continuing throughout the following, technologically remarkable centuries, astronomers invented telescopes, then bigger telescopes that could see fainter things. Then they added cameras and spectrographs that collected and dissected starlight, and later they refined the cameras' focus by flying satellites above Earth's distorting atmosphere. The technologies' outcome, like the number of stars, is also stupefying: The Mesopotamians might have been off by a finger or so held at arm's length, maybe a degree of arc, and Hipparchus was off by about half a degree, or 30 arc minutes. The Gaia satellite, however, is off by no more than 24 millionths of an arc second, the width of a human hair from 1,000 kilometers away. This kind of precision means astronomers can find structures in the Milky Way that are not only features of its map but also evidence of its history. Among the first structures found were confirmations of stars arcing through the halo in streams that were born together and still travel as one.
Ana Bonaca recently got her first faculty job, as a staff scientist at the Carnegie Observatories in Pasadena, Calif., but she's been interested in our galaxy since she was in middle school. She learned to use the floods of SDSS data to look for structures in the halo of stars that surrounds and is bound to the Milky Way. The halo, at the galaxy's farthest reaches, was known to be made of old stars and assumed to be featureless, but it is so dim that astronomers knew little else. “I was really drawn to the large-data-set, needle-in-a-haystack aspect” of the work, Bonaca says. In graduate school, her adviser suggested she look in the halo specifically for stellar streams. She didn't know what stellar streams were but soon decided that they were “pretty cool.”
In 2006 first the SDSS and then other surveys began verifying halo stars with the same colors and brightnesses that seemed to move together in long streams, like one that Bonaca later worked on called Triangulum. Astronomers suspected the streams came from outside the galaxy—that their stars had been born together in some little, nearby galaxy. Then they were pulled into a stream when that galaxy came too close to the gravitational tides of the much larger Milky Way.
This picture made sense, but verifying it was complicated. First, to believe that stars were in a stream, astronomers needed to see that the stars were related—that they'd been born in the same galaxy and were the same age. When stars are born together within the same gas cloud, they bear the distinct chemical signatures of the elements present in the cloud. As stars age, they turn light elements into heavier ones, which astronomers call “metals,” then die in explosions that scatter the metals back into the gas around them. The more generations of stars that have lived and died in a galaxy's gas, the more metal-rich are the new stars born inside it; the more metal-rich, the younger the star. Stars in a stream, formed in gas clouds in the same galaxy, should have the same chemistry and ages.
Second, stars in a stream should share the same motions. Motions toward or away from us are easy to calculate from stars' spectra, but measurements of their so-called proper motions across the sky have been imprecise. “If your error bars are too large, you can't see the stream,” says Amina Helmi of the University of Groningen in the Netherlands, for whom the Helmi stream was named. “We were desperately waiting for Gaia.”
In 2016 Gaia began releasing its wealth of data—chemical compositions, ages, and precise three-dimensional locations and motions, including proper motions—for billions of stars. With Gaia data plus measurements from other surveys, notably SDSS's APOGEE, astronomers were able to reliably identify which stars were born outside the Milky Way and had immigrated in and which had been born here, “in situ.” They could not only verify foreign star streams but also track each stream's orbit back to its own little galaxy.
Credit: Ron Miller (Milky Way side view); S. Payne-Wardenaar/K. Malhan, MPIA (merger illustration)
By 2021 astronomers had found 60 streams in the halo; 23 of them had likely birthplaces in dwarf galaxies or in the Milky Way's globular clusters (mysterious bound balls of up to a million stars that orbit our galaxy). Altogether, Bonaca says, there were “10 times more streams than before Gaia.” The stars are generally around 10 billion years old. The ages of the streams themselves are harder to estimate, but they are probably a few billion years old. Bonaca expects that astronomers will eventually find around 100 streams.
The streams running through the halo were some of the first signs of the galaxy's departure from stability. Then scientists began uncovering other groupings of stars that didn't follow expected patterns. In 2017 Bonaca and her team found a batch of Milky Way stars in the wrong place: they were in the old, metal-poor halo and had the orbits of old halo stars, but they had the metal-rich chemistry of younger stars from the Milky Way's disk. Bonaca wondered whether they were disk stars that had somehow wandered up into the halo.
The next year a team led by Vasily Belokurov of the University of Cambridge found an entirely different batch of stars in the halo that were going unusually fast and in the opposite direction from the rest of the halo. They named the wrong-way batch, which was bean-shaped, Sausage. A different team, led by Helmi, found that the bean's stars were also old and metal-poor; they called the bean Gaia-Enceladus, for the earth goddess Gaia's son, Enceladus. And in 2023 Bonaca and her colleagues found a stream of stars with the same old, metal-poor chemistry and wrong-way motion as the bean and thought this stream was probably tracing the bean's fall into the Milky Way. The astronomical community pragmatically settled on a compromise name for the bean, the Gaia-Enceladus Sausage (GES); the generic noun for a GES-type entity is “blob.”
Meanwhile the Belokurov team had “rediscovered” Bonaca's team's misplaced disk stars, by now known to be part of the GES. In other words, in the midst of a foreign blob of metal-poor stars was a group of metal-rich stars native to the Milky Way. He and his colleagues suggested that when the GES collided with our galaxy, it splashed the native stars out of their normal orbits in the disk and up into the halo. They called the star group Splash.
Putting their blobs, streams and splashes together, astronomers concluded that between eight billion and 10 billion years ago, Enceladus—about a quarter of the size of the Milky Way—struck our galaxy head-on and merged into it as a blob. “Head-on, you smash in, fall apart fast and die,” Belokurov says. GES stars now make up most of the Milky Way's halo, and the merger thickened its disk. Bonaca calls it “the most transformative event in Milky Way history.”
Older, less violent transformations had happened not in the halo but in the body of the galaxy itself. In 2022 three different teams found signs of a protogalaxy apparently turning into a galaxy. Again, verification was complicated and hinged on knowing which stars were native to the Milky Way.
Harvard's Conroy was part of a team that measured the in situ stars' chemistry and found two populations: one group was ancient, metal-poor, moving chaotically and forming stars slowly; the other was younger, metal-rich, moving coherently and forming stars 10 times faster. The astronomers thought the populations represented different stages of galactic history and called these stages “simmering” and “boiling,” respectively. Meanwhile Belokurov and a team measured in situ stars' orbits and also found two epochs, he says: an early one with metal-poor stars' orbits going “all over the place” and a later one with stars richer in metals that were orbiting more coherently—“a transition,” he says, “from hot mess to relatively cold spinning disk.” They called the hot mess Aurora, after the ancient Greek goddess of dawn. Hans-Walter Rix of the Max Planck Institute for Astronomy in Heidelberg, Germany, and his team looked at the chemistry of two million in situ stars across the sky and found a gravitationally bound group of ancient, metal-poor stars in the center of the galaxy. They called it “the poor old heart” of the Milky Way.
Names aside, all three teams agree that they're probably studying the same transformation: a chaotic protogalaxy full of old, metal-poor stars going in no particular direction that then spun up into a disk and began to form new stars like fireworks. Bonaca, who was on Conroy's team, isn't sure the observations have converged into one consistent story, “but it does look like we're seeing some of the same things,” she says. “It's a little like the elephant.”
Stars tell only part of the story because the Milky Way is only partly stars—the rest is mostly gas. Stars are born from gas clouds, so the two are intimately related. Nevertheless, astronomers who study stars and those who study gas work in largely nonoverlapping communities. Benjamin belongs in both but identifies more with the gas people than with the star people. Because stars are born in gas and later enrich that gas with the elements they produce, gas astronomers are interested in how the galaxy stays alive and therefore its present. And because stars retain the orbits and chemistry of their origins, star people tend to be interested in how the galaxy evolved and therefore its past. “I think of a galaxy as alive and breathing,” Benjamin says, “and those [star] folks treat it like a crime scene that needs forensics.”
Astronomers have been able to map gas clouds for only the past 100 years or so, because the clouds—which are large, diffuse and dim—are hard to study. Observers could outline their positions in the sky but could only approximate their distances and shapes. Gaia data allow scientists to detect gas clouds through their stars, but the method is indirect, done via a proxy of a proxy.
Gas clouds are 99 percent gas; the other 1 percent is dust, a fine soot mixed with the gas so thoroughly that a map of the dust is more or less a map of the gas. Dust can be identified by its effect on starlight: stars shining through dust look redder and dimmer. By mapping reddened, dimmed stars, scientists can trace an outline of the dust and therefore the gas. The dust-filled gas clouds are also peppered with well-known and precisely located stars, and astronomers can connect these stellar dots to map out the clouds. Still, a measurement this indirect, says João Alves of the University of Vienna, is like describing the elephant by “touching the hair on its tail” and looking at “one part in a million of the elephant.”
A team of astronomers led by Catherine Zucker at the Center for Astrophysics | Harvard & Smithsonian found a dozen or more long, threadlike clouds of gas, scattered like toothpicks throughout the galaxy's spiral arms, that might serve as birthplaces for the arms' wealth of new stars; the discoverers call the clouds “bones.” Another team uncovered a single, much larger but similarly long and narrow gas cloud that it calls the Split. And a third group, led by Alves, mapped the gas clouds that had clusters of newborn stars—the “local stellar nurseries,” Alves says. “The shock was that the nurseries are all aligned in a narrow line.” Seen from the side, this alignment looks like a wave that, like the Split but larger, undulates through the plane of the galaxy; the researchers named it the Radcliffe Wave. The Radcliffe Wave is 10 times longer and 100 times wider than the bones.
One reason these filaments of gas are interesting is that they—the bones especially—probably coincide with the galaxy's spiral arms, and no one yet knows how many arms the galaxy has. So far the arms look less like coherent structures than like arms plus branching feathers, making a count of their number dicey. If we could look at our galaxy from the outside, we would probably see it as having something between the disorganized, blotchy arms of a so-called flocculent spiral and the elegant, orderly arms of a “grand design” spiral. The consensus: spiral arms are best studied in galaxies we don't live in.
More recently, another team led by Zucker mapped what is known as the Local Bubble, a nearly empty region around the solar system made of hot, rarefied gas, and found the bubble outlined by groups of young stars, all moving outward. The researchers proposed that the bubble was created about 14 million years ago when a cluster of stars exploded as supernovae, sweeping up ambient gas and carrying it into a large sphere on whose surface the gas cooled into clouds and began forming its own stars.
Benjamin and others wonder whether the gas structures—the bones, the Split, the Radcliffe Wave and the Local Bubble—are variants of the same thing: long filaments of gas inside which smaller clouds are compressed into stars. “You see this long, dark, [dusty] thing,” Benjamin says, “and then, boom! There's a little bright bubble forming inside it, and then you see more dark line and then another bright bubble.” It's “like pearls on a necklace,” Alves says.
And maybe the Split, Radcliffe Wave and Local Bubble are historically related. The Local Bubble lies between the Split and the Radcliffe Wave. “We live in a bubble between a big snake and a smaller one,” Alves says. He and his teammates speculate that if we could rewind time to see the locations and motions of the Split and the Radcliffe Wave 15 million years ago, we'd find that the two were close enough to intersect. Right at their presumed crossing point, where gas would have been densest and mostly likely to produce new stars, astronomers see a lively crowd of young stars in a group of clusters called the Scorpius-Centaurus association, Sco-Cen for short. Moreover, the Split-Wave intersection and Sco-Cen happen to be at the center of the Local Bubble and therefore arguably the bubble's origin. “But this is still not for sure,” Alves says. “It just makes all the sense that [the intersection is] where the gas came from to form Sco-Cen.”
If the story told by the stars is the galaxy's history of assembling itself, and if the story told by the gas is the galaxy's cycles of star formation, then the stars and gas together should show the galaxy's past and present, a movie that reveals what Benjamin calls “evolving disequilibrium.”
Here is the elephant so far: Thirteen billion years ago, in a universe that was then less than a billion years old, the Milky Way was born as a shapeless cloud of gas and dust, forming metal-poor stars and rotating incoherently so that its stars' orbits were also haphazard. For the first billion or so years, smaller clouds and dwarf galaxies crashed into the baby Milky Way, sending up sprays of both immigrant and native stars into a halo. Gas carried by incoming colliders also set off more star formation in the Milky Way.
By about 12.5 billion years ago the galaxy was rotating more coherently; one billion to two billion years later it had spun up into a disk in which stars' orbits were tidily circular. Stars now formed at a quiet simmer, burned quickly through their lives and died explosively, enriching the gas out of which the next generations of increasingly metal-rich stars would be born.
Ten or so billion years ago the Enceladus galaxy collided with the Milky Way and, over the next two billion years, dissolved into it. The Gaia Enceladus Sausage took over the halo, sped up the stars in the Milky Way's thick disk and poured in gas, which, added to the Milky Way's gas, increased star formation. Gradually over the next two billion years inside the thick disk, gas and stars settled out into a denser, thin disk and collected into spiral arms.
Beginning around six billion years ago, a dwarf galaxy named Sagittarius sideswiped the Milky Way and swung around it. Every few hundred million years after that it brushed past the Milky Way again, each time “leaking stars in a trail,” Belokurov says, creating streams that curved through the Milky Way's halo, wrapping around it twice. During the next five billion or so years, other incoming objects did the same until the entire Milky Way was surrounded with streamers. By then, in the spiral arms of the concentrated thin disk, gas had gathered into long threads—bones, waves, splits, filaments—along which stars lit up in clusters.
Closer to the sun, starting about 15 million years ago, massive stars in the Sco-Cen association formed, lived their fast lives and blew up, creating the Local Bubble, on whose dense surface more stars formed. The 37 clusters that now make the Sco-Cen association have fired off in bursts roughly every five million years, carving out more bubbles with more dense surfaces forming more stars, ensuring that the galactic neighborhood foams with new sparklers. The filamentary clouds don't survive the floods of radiation from star birth, and after five million to 20 million years they have “sheared apart” back into the galaxy, Benjamin says. There the gas will eventually cool and, under the influence of gravity and rotation, recondense into filaments and then again into stars.
And on planet Earth, maybe 4,000 years ago, a baby born in Mesopotamia grew up to know the names of the stars and constellations as though they were family or gods, to write them down in stone and use them to plant crops, measure time and predict lives. We've renamed the constellations since the god Marduk set them up: the Bull of Heaven is our Taurus; Scorpion is our Scorpio; Pabilsag is Sagittarius; and the True Shepherd of Heaven is now known as Orion. But we still use the constellations to locate ourselves in the galaxy, and we name its places, its clouds and streams, for earthly analogues.
With the new surveys' maps, we can see how constellations have warped and shifted with time and how the galaxy has and will continue to change. “We can run the movie forward and backward,” Benjamin says. “We can do that with certainty.”
The gas and star maps are complete near the sun but get hazy farther out. By 2023 astronomers had still mapped only about two billion of the Milky Way's 100 billion stars. “If the sun is my nose, we're still, like, here,” says Alyssa Goodman of the Center for Astrophysics | Harvard & Smithsonian, touching her hands to either side of her face. “And the scale of the galaxy is, like, way out beyond the end of my arms. And so we're just trying to get, like, here, here, here.” With each “here” she moves her hands farther and farther out until her arms are wide open, measuring the scale of the Milky Way with her human body, taking the galaxy as personally as any Mesopotamian baby.