Once we are settled in a red pickup truck, my driver and I leave the airport behind and make our way through Chile's Atacama Desert toward an isolated peak known as Cerro Manqui. Two hours later, as the car hugs a curve of the winding road that summits the mountain, I welcome a familiar sight: sunlight bouncing off the silver shells of the twin Magellan telescopes, Baade and Clay. My heart beats a little faster. Starting tomorrow night, the Clay telescope is all mine.

I travel from Boston to Las Campanas Observatory about three times every year to help unravel some of the remaining mysteries of the Milky Way's evolution. Astronomers are intimately familiar with the anatomy of our galaxy, but we still do not know all the details of its birth and development. Computer simulations of the early universe suggest that thousands of small galaxies once surrounded the young Milky Way, which grew larger by consuming many of its smaller brethren. To help determine whether these simulations are correct, I compare the chemical composition of ancient stars in the outskirts of our galaxy—a region known as the halo—with that of ancient stars in dwarf galaxies that still orbit the Milky Way today. If the simulations are right, then ancient halo stars and dwarf galaxy stars should be made from the same stuff.

Over the past several years that is exactly what chemical analysis has revealed. In all likelihood, the Milky Way has expanded by gobbling up dwarf galaxies and incorporating their stars into its halo. Even now our galaxy seems to be fattening itself with stellar streams torn from galactic neighbors. Astronomers, however, have not yet collected enough data to write these ideas into the textbooks. Like every good observer, I am always in search of more evidence. Every now and then an astronomer needs to leave the office and university behind, travel somewhere remote, away from any urban hullabaloo—preferably, somewhere with high elevation—and confront the night sky in all its naked beauty. On journeys like this, I remember why I fell in love with science in the first place. That is what I would like to show you.


As usual, I have arrived at the observatory a day before my turn with the telescope so that I have time to adopt the nocturnal schedule my research requires. At this time of year, a typical workday lasts from 3 p.m. to 6 a.m., with the night's observations starting around 6 p.m. I rest for an hour before having dinner with other astronomers in the lodge, where the geekiness is palpable. We talk about recent studies we have found interesting, any technical problems people have had with the telescopes, and the weather forecast—everyone dreads cloudy skies.

After dinner, I visit the operators and technical staff who maintain the Clay optical telescope and its impressive 21-foot-diameter mirror. Only relatively large telescopes like this one are capable of collecting sufficient light from the dim and distant stars I study. Even though I am not observing tonight, I like to talk with the staff and the current observer and learn what has happened at Las Campanas since my last visit.

Around 2 a.m.—having stayed awake long enough to begin shifting my sleep cycle—I leave the telescope and step into the cool night air, where I find my way among shrubs and stones. The Atacama, the driest desert in the world, is the ideal place to study stars: there is almost no water in the air to bend beams of starlight away from the telescopes. The Southern Hemisphere offers unparalleled views of the Milky Way even without a telescope. I tilt my head back and stare into the center of our galaxy, where countless stars are scattered like flecks of diamond in molasses.

If you were to peer at the Milky Way edgewise, it would look like an egg, sunny-side up: a bright, dense yolk of stars called the galactic center, around which the galaxy's spirals form a thin saucer known as the galactic disk. An evanescent halo of old stars envelops the entire galactic disk. About 30 known dwarf galaxies spin through the outermost regions of the halo. On average, typical dwarf galaxies contain only a few billion stars, far fewer than the 200 billion to 400 billion stars in the comparatively gigantic Milky Way. Some particularly dim dwarf galaxies may contain only thousands of stars, although it is difficult to count stars in such faint clusters.

My research primarily focuses on stars in ultrafaint dwarf galaxies that astronomers spotted only in the past 10 years. Stars in these galaxies seem to be some of the oldest ever discovered. We know these stars are old because of the proportions of chemical elements they contain. After the big bang, the first stars in the universe formed from gaseous clouds of hydrogen, helium and tiny traces of lithium—the lightest of all elements and the only ones that existed at the time. As those first stars aged, the nuclear reactions in their cores produced heavier elements such as carbon, oxygen, nitrogen and iron, which spewed into space when these stars exploded as supernovae. A new generation of stars formed from gaseous clouds enriched by these heavier elements, which, along with lithium, astronomers call “metals” for convenience. Only stars that formed in later generations contain substantial amounts of metals. I study metal-poor stars that were born in the universe's infancy. Ultrafaint dwarf galaxies have fewer stars than their more luminous peers, but they have a higher proportion of metal-deficient stars—they are most likely relics from a time long past.

I walk from the telescope to the lodge guided by starlight alone—no need for a flashlight. Just me and the stars.

All in a Night's Work

After sleeping through most of my second day, I prepare for my first night of observation with the telescope. I take my seat at the observer's workplace—a desk on which are a few computer screens that tell me about the condition of the telescope, the weather and the positions of stars. The telescope operator, who maneuvers the instrument on my command, sits in front of a wall of 15 screens arranged in several rows.

The week before I arrived in Chile, I made a “target list” of dwarf galaxy stars, ordered by priority. After reviewing the weather conditions, I choose the first star on the list, ask the operator to move the telescope into position and begin collecting starlight.

The ribbons of starlight that travel from dwarf galaxies some 130,000 light-years away carry the stars' chemical DNA—but the code must be deciphered. The Clay telescope is equipped with a high-resolution spectrograph that stretches the beam of starlight into a rainbow of different wavelengths, which I view on a small computer screen. Slicing through the rainbow at different points are black vertical bars known as absorption lines, which correspond to the abundances of different chemical elements in the outer shell of the star. The thinner the absorption line, the less of that particular element exists in the star. In fact, high-resolution spectroscopy is precise enough to tell me how many individual atoms of each chemical element a star contains.

All the starlight I have collected and analyzed in the past several years shows me that both halo stars and dim dwarf galaxy stars have very weak absorption lines corresponding to heavy elements such as iron. In the Milk Way's halo, for example, I discovered the most iron-deficient star in the universe, which has only 1 percent as much iron as Earth's core. For comparison's sake, consider that this star is about 60 percent as massive as the sun, which is 300,000 times more massive than our planet.

Such metal-poor halo stars could not have been born in the Milky Way among relatively recent generations of stars. Rather they must have formed from the same kinds of gaseous clouds that birthed ancient dwarf galaxy stars—clouds that existed only in the universe's infancy, before stellar furnaces churned out the heavier elements. The evidence suggests that ancient halo stars are chemically similar to dwarf galaxy stars because they were once part of dwarf galaxies, too. Over time the Milky Way ingested these nearby dwarf galaxies, stealing their stars and growing larger all the while. Yet chemical analysis is not the only evidence of our galaxy's cannibalism. Astronomers have also found what we think are the stains of former meals—stellar streams in the halo, which were likely unspooled from satellite galaxies caught in the Milky Way's gravitational field. Right now the Milky Way is eating up the Sagittarius dwarf elliptical galaxy bit by bit as the satellite zips in arcs around our galaxy. With every turn, stars are torn away from Sagittarius and flung into our galaxy's halo.

Around 7 a.m., more than 12 hours since I first entered the telescope observation room, I am satisfied with the data I have collected on the first stars in my target list. Time to call it a night. I gather my notes, leave the telescope and make the short journey down the mountain to my bedroom in the lodge. Already I am imagining myself drawing the thick, sun-proof shades on my window and resting my head against my pillow. The morning twilight cloaks the stars overhead, but I know they are there—burning as they have for billions of years.

Further Observations

I drag myself out of bed at 3 p.m. and, after some dinner, prepare to make more observations with the telescope. I cannot afford to waste a single minute, especially considering that each evening of observation costs more than $50,000, so I plan my nights carefully.

Whenever I observe a star, I need to collect a sufficient number of photons to later make a meaningful analysis of that star's chemical composition. The fainter the star, the more time I need to collect enough photons. Ideally, I want to observe each dwarf galaxy star on my target list for a total of 10 hours because these stars are so faint—halo stars, in contrast, require only one to three hours of exposure. As Earth rotates around its own axis, however, Las Campanas turns away from the region of space I am studying, making it impossible to observe any of my target dwarf galaxy stars for more than four or five hours a night. To compensate, I observe the same set of stars over the course of several nights. There is another complication: high-energy cosmic rays constantly bombard the planet—hitting the telescope's detector and degrading the data. I have found that an efficient way to strike the right balance between collecting enough starlight but not too many cosmic rays is to break up my observations into 55-minute chunks. Shorter than 55 minutes, and I will have not collected enough photons; too much longer than 55 minutes, and the instruments will have been hit by too many cosmic rays. I usually observe one star for four or five 55-minute chunks and move on to the next star in a different part of the sky.

When it is time to switch from observing one star to another, I must carefully review all the data available to me: the number of photons I have collected so far, the positions of my target stars in the night sky, and the weather forecast. The telescope operator is waiting for my decision. Let's say, for example, that I have not collected as many photons as I would like from the first star I was observing but that this star will soon disappear below the horizon. I need to decide whether to stick with the star a little while longer or move on to another and hope that the skies will be clear enough to observe the first star again another night. If I am lucky, I am able to scamper downstairs to the kitchen to make myself a sandwich, but for most of the time I am glued to my computer screens until I have collected enough photons to call it a night.

A Change in the Weather

Around 6:30 p.m., I step onto the catwalk outside the Clay telescope before a new night of observation. Watching the sunset at Las Campanas is something of a ritual. The sun sinks slowly below the horizon, draping the hilltops in veils of pink and peach. Each sunset marks a new night of observation—as long as the weather obliges. My third night in the telescope begins well enough, but before long I am frowning at the weather reports on the monitors in front of me. I open the telescope door and stick my head into the night air. Clouds thicker than clotted cream have crowded the Cerro Manqui peak. There is not much that can be done. I won't be observing any more stars tonight. I sit at my laptop and answer e-mails I have ignored for too long, sort through data from previous studies, and write—in fact, I wrote most of this article that cloudy night.

When I take a break from my writing, images of yet undiscovered dwarf galaxies swim through my mind. Computer simulations of our galaxy's birth suggest there are many more dwarf galaxies orbiting the Milky Way than we have discovered so far. We have mapped all the bright dwarf galaxies. The ones we do not know about yet are either much fainter or farther away, which means we need an especially keen eye to find them. The Carnegie Institution for Science plans to build a new telescope at Las Campanas, on a hill that neighbors the Cerro Manqui peak—an instrument boasting an 82-foot-diameter mirror. That is nearly four times the diameter of the mirror I use now. With its giant mirror and accompanying spectrograph, the new telescope will let me gaze into far-flung regions of the Milky Way's halo, where I hope to find more metal-deficient stars. The more observations we make, the closer we get to filling in all the gaps in the story of our galaxy and of how the Milky Way became what it is today.