This article originally appeared in The New York Times Magazine and is reprinted on Longform by permission of the author.
LATE ON THE MORNING of April 3, the expedition ship Alucia rocked violently on the South Atlantic Ocean in the middle of a squall. On the aft deck, the crew huddled together in rain slickers and gazed across the heaving seas to a yellow blur on the horizon. This was an unmanned reconnaissance submarine carrying 15,000 photographs that they were nearly desperate to see. But it had buoyed to the surface just as the squall sprang up, and with 30-knot winds and four-foot swells that splashed over the stern, it was too dangerous to retrieve the sub. So they watched and waited.
For eight days, the Alucia had been trolling the ocean near a spot known as the L.K.P., or the Last Known Position of Flight 447, the Air France jet that vanished in June 2009, about halfway between South America and Africa. In the nearly two years since, three other search teams went looking for the wreckage, but this was the Alucia’s first try. The ship carried three Remus 6000 submarines, some of the most advanced underwater search vehicles on earth, which swept the seafloor in 20-hour runs, then surfaced to deliver sonar imagery to the Alucia’s scientific team, who pored over the data in 12-hour shifts around the clock. So far, they had not found the plane, but the day before, one scientist pointed at something unusual on the monitor and said, “What about this?” And ever since, the air on the Alucia was charged.
Everyone knew the stakes. This wasn’t a scan of the Sargasso Sea or a study of salinity samples. The families of 228 passengers were restless for results. The search had already taken two years and cost more than $25 million. Another $12 million was committed to the Alucia this year, but French investigators had quietly decided that this year would be the last. If the Alucia did not find the plane, no one ever would.
As expedition leader, Michael Purcell was equal parts colleague and boss, with a raspy voice and a sonic laugh and a playful sarcasm, but he knew the Remus subs as well as anyone. Looking at the fuzzy mark on the monitor, he knew they had found something unnatural. It was too long and straight to be geologic. It was unlike anything else on the seafloor. On the other hand, if it wasn’t Flight 447, Purcell knew the disappointment would be palpable. As he prepared the photographic sub to return to the bottom for an 18-hour mission, Purcell whispered to another scientist, “I’m 95 percent sure that’s it, but man, if it’s not, it’s going to be a long two and a half months.” The sub went down at 9:45 p.m. At 2 a.m., Purcell was still awake in his cabin. He picked up his journal. “Tired but not sleepy,” he wrote. “May have found the plane today. Everyone is on edge.”
Four hours later, Purcell was up with the sun, and by late morning he was on deck with the crew, watching the Remus bob in the distance. A little after 1 p.m., they pulled the sub onboard and attached two thick cables to upload its data into the computers in the mission-control room. They drew the curtains around the room, so nonscientific crew members could not see in, and yanked the satellite uplink offline, so no one could leak the news. Then they crowded around the computer monitor as the first images of Flight 447 came onscreen: engines, landing gear and sections of fuselage, all unmistakably vivid on the ocean floor. But as they turned the satellite back on and began sending the first photos to air-crash investigators in France, the deeper implications of their discovery were just beginning to surface.
THE VANISHING of flight 447 was easy to bend into myth. No other passenger jet in modern history had disappeared so completely — without a Mayday call or a witness or even a trace on radar. The airplane itself, an Airbus A330, was considered to be among the safest. It was equipped with the automated fly-by-wire system, which is designed to reduce human error by letting computers control many aspects of the flight. And when, in the middle of the night, in the middle of the ocean, Flight 447 seemed to disappear from the sky, it was tempting to deliver a tidy narrative about the hubris of building a self-flying airplane, Icarus falling from the sky. Or maybe Flight 447 was the Titanic, an uncrashable ship at the bottom of the sea.
But the mystery of Flight 447 demands real answers, and not only for families of the dead. Every airplane, like every car, is built from thousands of parts made by dozens of manufacturers, and many of those parts appear in multiple aircraft. Until we know which parts, if any, failed on Flight 447, it is impossible to know which other planes might be at risk. As Jean-Paul Troadec, the director the French Bureau of Inquiry and Analysis (B.E.A.), which is investigating the crash, told me: “If there is a technical problem on this aircraft, we have to know. But this is not just an Airbus question. This accident could apply to every aircraft.”
Over the last two years, several theories have emerged to explain the crash, but there is still no clear consensus. Early reports suggested terrorism — two names on the passenger list seemed to match those of Islamic radicals — but this was quickly ruled out. Weeks after the crash, as thousands of pieces of the airframe were scooped from the ocean surface and collected in a French government warehouse in Toulouse, some aviation experts speculated that the plane may have come apart in the air; others, looking at the same evidence, insisted that it landed in one piece and then shattered. There are mysteries about the plane’s flight path as well. Just before Flight 447 went down, it charged into a massive cluster of roiling clouds, even as three other planes made wide turns to avoid the weather. Why Flight 447 flew into the clouds may help explain why it never flew out.
In the hours before the flight took off from Rio de Janeiro, there were no signs of mechanical trouble. Pilots who flew the plane in from Paris just a few hours earlier had reported problems with a radio panel on the left side of the cockpit, but mechanics swapped that panel, and there was a duplicate on the other side. Flight 447 was, or seemed, perfectly safe to fly.
A little after 6 p.m., the passengers filed onto the plane, the cabin door closed and Flight 447 pushed back from the gate. The flight has since been renumbered Flight 445 but otherwise remains the same. You barrel down the runway on an Airbus A330, then sweep up over the water, the lights of Rio shrinking to a yellow haze around the city’s black lagoon as you climb toward 35,000 feet, coursing north along the edge of the continent before turning east to Africa.
All flights over the ocean cross certain checkpoints, and the most critical one for Flight 447 was Tasil Point. Almost halfway between South America and Africa, it serves as a switching station for air traffic control. On one side of Tasil, pilots report to Brazil; on the other, they belong to Senegal. In theory, it’s a place where air traffic controllers on both sides are watching; but in practice, it’s often a dark spot on radar, too far away from either continent to see. It’s also a place where high-frequency radios often cannot be heard, and it falls on top of the meteorological equator, where the winds of the hemispheres collide. Some days they die, and the stillness can be transfixing — sailors call the region the doldrums. But other times, the winds whip together into a black anvil storm that blooms through the depths of the troposphere, with a luminous violet glow gathering around ship masts and airplane wings — sailors call this St. Elmo’s Fire.
As Flight 447 began its journey toward Tasil Point, the skies seemed reassuringly clear. In the cockpit, the pilots chattered with Brazilian air traffic control, calling out altitude and radio frequencies. Sometime around midnight, the waxing moon, which had been gleaming through the port-side windows, dropped below the horizon, and Flight 447 was alone in the sky.
At 1:35, the pilots called Brazil to read off their altitude and flight plan. Three seconds later, controllers called back to ask when they would reach Tasil Point. Seven seconds passed, and Brazil called again. Another six seconds, and again. Flight 447 was gone.
TO A PILOT, the distant nowhere of Tasil Point is a treacherous cone of radio silence, but to an oceanographer like Purcell, there are few better places to be. The bottom of the ocean, after all, is not just a great expanse of sand; it is covered with hills and steppes and valleys, each with its own unfathomable intricacies, and among these the underwater mountains of the midocean ridge are among the most mysterious places on earth.
The mountains date to the beginning of the planet as we know it today. Two hundred and fifty million years ago, as the supercontinent Pangaea broke into the seven continents, the stretching landmass tore fissures in the crust of the world, gaping canyons that spread apart and filled with the modern oceans. But under the water, volcanoes continued to erupt, sending up magma and forming mountains of basalt. Today, that mountain range still zigzags down the Atlantic and across the South Pacific and through the lower Indian Ocean, wrapping the globe like a crudely stitched soccer ball. Together they make the largest mountain range on the planet, 50,000 miles long and almost entirely unexplored, as tall as the Andes but two miles deep.
Until the last few decades, the tools to explore these underwater mountains did not exist. Early submarines could go deep enough, but the terrain was impossibly forbidding. Even today, the machinery for a journey to the midocean ridge is prohibitively expensive for most teams. In the oceanography world, it is common to spend more than $1 million on an expedition, but a trip to the mountains at Tasil Point can easily cost 10 times as much and require submarines so advanced that only a handful of scientists know how to use them.
In the weeks after Flight 447 went down, as the French and Brazilian navies trolled the ocean near Tasil Point, they found more than 3,000 pieces of debris scattered across the surface, including fragments of the wings, most of the tail fin and the seats of cabin crew. They also found the bodies of 50 passengers, some still clothed and some still wearing jewelry. But as they transported the bodies to a morgue in Brazil and assembled the wreckage at a warehouse in Toulouse, the glimmering surface of the ocean remained as impenetrable as a shield of diamonds. When French investigators began calling oceanographers and deep-sea adventurers in a desperate effort to reach the bottom, the name of one organization kept coming up.
The Woods Hole Oceanographic Institution is a sprawling complex of shingled houses, brick laboratories and cavernous warehouses spread across two campuses in southern Cape Cod. In the 1970s, Woods Hole scientists discovered hydrothermal vents brimming with life at the bottom of the Pacific. In the 1980s, they discovered the Titanic under more than 12,000 feet of water. In the 1990s, their submarine laboratory pushed boundaries in underwater technology, building the first Remus in ’95, then adding cameras and new sensors and smarter computers, culminating in the Remus 6000. No other organization had more experience with the Remus than the scientists like Purcell who built it; and the director of special projects at Woods Hole, David Gallo, is an expert in the underwater mountains. In fact, he wrote his dissertation on the midocean ridge. When I asked Michel Guerard, a vice president at Airbus, how Woods Hole was selected to lead the search, he shrugged and said, “Because no one else in the world can do it.”
To find one of the most advanced airplanes in the sky, they would have to travel to one of the most primitive places on earth — and seek out one of the most archaic devices in modern aviation, a relic that has barely changed in the last half-century of flight: the black box.
“IT'S RIDICULOUS," Peter Goelz says. “There is absolutely no reason not to have live-streaming data.” As managing director of the National Transportation Safety Board in the late 1990s, Goelz saw his share of accidents, but the disappearance of Flight 447 got under his skin. The fact that in an era of wireless technology, when passengers on some jets can surf the Internet, the most valuable information about a flight is still stored on the very airplane that has, by definition, crashed before the information is needed, did not sit well with Goelz.
“I’ve had arguments with pilots over this,” he told me. “They don’t want anyone monitoring their performance. But every 7-Eleven clerk in America has a bank of cameras on him. Why can’t pilots?”
The technology to stream data from planes does have some limitations. To stream all the data from every plane might well require more satellite bandwidth than the total amount in orbit. But some companies have found a solution: instead of streaming all data, they equip planes with a system that streams on demand. In case of trouble, a pilot pushes a button to begin sending down data, and in some conditions, like the failure of autopilot, the data streams automatically. Airlines could also simply stream the data from select flights, like those crossing rough terrain — the Himalayas, for example, or the midocean ridge. When I mentioned that idea to Alain Bassil, the chief operating officer of Air France, he nodded and said, “That would be a good solution.” Then he added, “Hopefully soon.”
In late April, after analyzing pictures from the Alucia, French investigators returned to Tasil Point and deployed a recovery submarine to a spot that appeared to hold one of the black boxes. Sure enough, the box was there — but its data was gone. At some point, the cylinder that holds the data apparently broke off. On May 1, the cylinder was found on the ocean floor nearby; and on May 3 the other black box’s cylinder, which contains the cockpit voice recording, was also located and recovered. In the days ahead, both cylinders will be delivered to the basement of the B.E.A. in Paris, where investigators may still be able to decipher them, even after two years underwater.
But even without the black-box data, several things are clear. From the meteorological charts near Tasil Point, the safety record of certain Airbus parts, the flight protocols of Air France and the air traffic control logs, it is possible to piece together some events from the fall of Flight 447 that answer some pivotal questions and raise others.
In the last four minutes before the crash, the airplane sent a series of 24 automatic fault messages to a maintenance center in France. Among these, the first to jump out at experts involved a part called the pitot probe. Pitots (pronounced PEE-toes) are small cylinders that sit outside the body of the plane to calculate airspeed. The cost of a pitot probe is not high — about $3,500 each for the model on Flight 447, which disappears in the $200 million cost of a plane — but their importance would be hard to overstate. Without them, a plane’s flight computer has no way to determine speed, and the automatic pilot shuts down. That means that if any of the pitot probes, sticking out into the wind, happen to get clogged with dirt or ice, the plane will suddenly revert to manual control, forcing pilots to take the stick of a half-million-pound aircraft in whatever conditions disrupted the pitot in the first place.
In theory, this shouldn’t cause a crash. The probes can be compared to a speedometer in a car: steady on the gas, and you’ll be fine. Pilots are trained to respond to pitot failure by maintaining pitch and thrust until the probes resume working. Most of the time, they do.
But during the period of manual control, the margin of error is thin. For a passenger jet like the A330, the ideal cruising speed is about 560 miles per hour. If you go much faster, the center of lift moves back on the wing, pushing the nose down and increasing velocity, until you soon approach the speed of sound. At that point, shockwaves develop on the wings, interrupting the flow of air and reducing lift. The nose of the plane then gets forced into a dive that the pilot may not be able to pull out of. Then again, if you go too slow, the airplane stalls and falls. A plane must maintain a minimum speed to generate lift, and the higher it travels, the faster it must go. At 35,000 feet, the gap between too fast and too slow narrows ever closer. Pilots call it coffin corner.
Looking over the recent history of pitot failure can be unnerving. Peter Goelz, the former N.T.S.B. director, recalled several episodes during his tenure, but one in particular stood out: the crash of Birgenair Flight 301 in February 1996. “We had a Boeing 757 that had been on the ground in the Dominican Republic for a month,” he said. The investigation concluded that “during that time, one of the pitot probes got an insect nest built in it. Well, the crew took off and flew the plane right into the ocean.” All 189 people onboard died.
The pitot probes on Flight 447 were even more vulnerable than most in conditions like those at Tasil Point. They were produced by a French company, Thales, and the model was known as AA. In the years leading up to the crash of Flight 447, the Thales AA was problematic in places where the meteorological conditions do funny things with water. At high altitude and low temperatures, water sometimes doesn’t freeze. Instead, it hovers, but as soon as something solid — like a pitot tube — flies through it, the water flash-freezes to form ice. Until heaters can melt the ice, the pitot probes are out.
This could happen to any kind of pitot probe, but by the summer of 2009, the problem of icing on the Thales AA was known to be especially common. Why the probes were still in use is a contentious question, but here is what we know for sure: Between 2003 and 2008, there were at least 17 cases in which the Thales AA had problems on the Airbus A330 and its sister plane, the A340. In September 2007, Airbus issued a “service bulletin” suggesting that airlines replace the AA pitots with a newer model, the BA, which was said to work better in ice.
In response, Air France’s official policy was to replace the AA pitots on its A330 planes “only when a failure occurred.” In August 2008, executives at Air France asked Airbus for proof that the BA pitots worked better in ice, and faced with the question, Airbus conceded that it did not have proof. So it removed the claim from the service bulletin. Another five months passed.
During that time, another airline, Air Caraïbes, experienced two close calls with the Thales AA on its Airbus A330s. The company’s chief executive immediately ordered the part scrapped from the fleet and alerted European regulators, who then began asking questions. In their conversations with Airbus, regulators learned of the 17 cases of icing, and they also discovered, looking at those cases, that the failures seemed to be happening more often (9 of the 17 occured in 2008). None of the failures seemed to signal an immediate danger, so the Thales AA was not removed from service. Regulators simply asked Airbus to watch the problem and report back in a year.
It wasn’t until April 2009 that Airbus delivered test results to Air France showing that the BA really did work better in ice. By then, 19 months had passed since the service bulletin suggesting the same thing, but now Air France made the change. At the end of April, the airline ordered replacement BA probes for its A330s, and on May 26, the first batch of probes arrived. Five days later, when Flight 447 took off in Rio, the probes were still in an Air France warehouse, and none of them had been installed. All three pitots on Flight 447 were the Thales AA.
THE HEADQUARTERS of Air France is a huge white box that squats near the runways at Charles de Gaulle Airport. One day this spring, I visited the building to meet with the company’s chief operating officer, Alain Bassil, in his office overlooking the tarmac. Bassil is a trim man in his mid-50s, with a tight smile and a mustache so neat that it seems penciled in.
Since the disappearance of Flight 447, some of the passengers’ families have come to regard Bassil and the other executives at Air France and Airbus as a cast of corporate villains. In Brazil, for example, I met with Maarten Van Sluys, a soft-spoken man in his late 40s who lost his sister, Adriana, on the flight and now leads the official association of Brazilian families. “There is a cover-up,” Van Sluys told me. “There is no doubt about it.” His claim was that French investigators located the wreckage long ago, already knew the cause of the crash and were keeping it secret to protect Air France and Airbus. “They don’t want to disclose problems with the fly-by-wire system,” he said.
This may sound extreme — and there is no evidence of a cover-up — but among the families of Flight 447, there is a range of opinion on the French investigation, and many of them express a sense of frustration. In Paris, I spoke with Gwenola Roger, whose boyfriend, Nicolas Toulliou, proposed to her a week before the crash on a moonlit walk by the Louvre. Sitting quietly in a friend’s apartment, Roger cut a vaguely regal figure as she spoke of her love for Nicolas and of her loss — then she lowered her voice and said: “They still haven’t said what they know. One day we will learn the truth.”
Over the last two years, the B.E.A. has in fact issued two bound reports, which include more than 200 pages of extensive data on the crash. From the first page, each report makes clear that “the investigation is not conducted in such a way as to apportion blame.” This is the formal mandate of the agency, which is expected to rise above the contentious aftermath of tragedy and deliver just the facts. All of which sounds very reasonable, until you are a grief-stricken family member, two years after the crash, holding the only official accident reports, which seem to announce on Page 1 that any wrongdoing will be ignored. “If you don’t want to find out who was at fault,” Van Sluys asked, “why do the investigation?”
In private, some B.E.A. investigators agree that they have found things that disturb them. After the plane’s final communication, for example, it took nearly 11 hours for a search team to be sent to Tasil Point. For the first hour, air traffic controllers generated a “virtual flight” on their computers, as is common practice, passing the plane along its intended route. For the next two hours, controllers checked periodically to see if anyone had seen the plane, and when a controller in Brazil asked a controller in Senegal if the plane had reached Cape Verde, the controller in Senegal said that Cape Verde hadn’t talked to them but not to worry; so the controller in Brazil didn’t. By the time Air France alerted a satellite search-and-rescue, 4 hours and 20 minutes had passed, and then it was another two hours before anyone notified the B.E.A. A search team lifted off in Dakar 10 hours after the last radio contact and for the next 45 minutes flew toward Cape Verde, where they assumed the plane had gone down.
When I asked the director of the B.E.A., Jean-Paul Troadec, if this was a suitable response time, he practically jumped from his seat and cried: “No! It’s not! The alert should have been much more quick!” Yet the reports from Troadec’s office draw no such conclusion. When I asked another B.E.A. investigator, Olivier Ferrante, whether it is difficult to write the reports without pointing out mistakes, he acknowledged that it is a matter of craft. “This requires discipline in report writing,” he said. “For example, we don’t use the word ‘fault.’ We prefer to use the word ‘error,’ which has more proactive connotations.”
Air-crash investigators at the N.T.S.B. say the American approach is very different. Jim Hall, a former N.T.S.B. chairman, told me that American investigators in the same position would have no trouble acknowledging if a search team took too long or if a plane was flying with faulty parts. “That would not be a problem at the N.T.S.B.,” he said. “If the board found that a part needed to be recalled, they would make that recommendation.” Hall explained that the difference in France is largely systemic: every crash is supposed to involve two parallel investigations, one by the B.E.A. to compile technical data and the other by a judge to consider liability. But when the judicial inquiry takes 22 months to begin, as it did for Flight 447, the half-scope of the B.E.A.’s work stands alone and incomplete. Hall told me that the French approach is “a mistake.” Goelz was less diplomatic. “There’s always been a sniff of politics at the B.E.A.,” he said, recalling the crash of Air France 4590, which erupted in flames just after takeoff at Charles de Gaulle in 2000. After a four-year probe, B.E.A. investigators decided that the crash was not caused by anything on the French plane but by a thin strip of metal that fell on the tarmac from a Continental flight minutes earlier. In Goelz’s opinion, “there was never any question that somebody else was going to be pointed out as the blame; and Continental became it.”
The French state is not just investigating Air France and Airbus but is also a prime investor in them. In fact, the French government, which nationalized Air France in 1945, currently owns nearly 16 percent of Air France-KLM, a stake worth about $830 million, and controls 3 of the 15 seats on the company’s board. The government also owns about 15 percent of the parent company for Airbus, which is worth another $3.8 billion. Of course, other countries also have a financial stake in their national airlines, and there is no evidence that the B.E.A. investigation has been compromised. But for a government to investigate a company it owns is the very definition of conflicted interest. It also turns out that the underwater search this spring was entirely financed by Air France and Airbus — and as one Air France executive told me, directly “by cash.” (Air France made a point of saying that it did not control the investigation and gave no directive to the search team, and this is confirmed by people at Woods Hole.) One day during the search, I asked Troadec what would happen if the Woods Hole team asked for additional funds for something the companies were unwilling to provide. Troadec sighed. “Any problem like that,” he assured me, “should be resolved in good faith.” When I asked Goelz if the N.T.S.B. would allow the target of an investigation to control the purse strings in the same way, he laughed. “No, no, no,” he said. “We would charge parties for underwater retrieval, but we would control the money.”
In the absence of a more critical public examination, Air France has conducted its own safety review and implemented its own improvements, though it has not explained what it has found or what it is changing or why. This is not to say that the airline hasn’t looked inward. In December 2009, five months after the crash, Air France commissioned an internal review of nearly all aspects of its operations. But when I asked Bassil and two other executives — Etienne Lichtenberger, the director of flight safety, and Bertrand Lebel, an executive vice president—– to explain what lessons Air France drew from the accident, they declined to say.
“The problem,” Bassil said, “is that we still don’t know what happened.”
“We do know certain facts,” I offered.
“There are very few facts,” Lebel said.
“There are two interim reports full of facts,” I said.
“We are not able to draw any conclusions,” Lebel said.
“You haven’t drawn any conclusions?” I asked.
“We are drawing conclusions,” he said, “and they will be explained.”
Bassil leaned forward. “We are not in a position to say anything else.”
“Can you explain why you’re not in a position to say?”
“Because we do not have anything else to say,” he said.
As I asked repeated questions about the pitot probes, the service bulletins and the search delays, they responded either in prepackaged talking points or by saying they had no comment.
On March 18, four days before the Alucia set sail, a French magistrate finally began the requisite judicial probe of the crash, and as that investigation makes its way toward court, it is joined by nearly a dozen civil lawsuits filed by family members. Perhaps as the legal discovery process moves forward, and if the black box data can be deciphered by analysts, new light will shine on the regulatory and corporate decisions that preceded to the crash. In the meantime, one of the only places in either B.E.A. report that offers any hint of criticism is directed toward the Brazilian morgue where the bodies of 50 passengers were taken to be identified. “At this stage of the investigation,” the report notes, “the B.E.A. has not had access to the autopsy data.”
ON A SUNDAY morning in mid-March, I met with Dr. Francisco Sarmento, the doctor who presided over the Flight 447 autopsies. This turned out to be a strange time to visit. Two days before I met with Sarmento, the morgue where the autopsies took place was shut down by inspectors, citing “blood on the walls,” “corpses stored on top of each other on shelves and on the floor,” “a strong stench of putrefaction” and a parade of other horrors, like a corpse “being dragged across the floor by two employees.” (The morgue has since reopened.)
Sarmento’s office in Recife turned out to be only somewhat more presentable. The floors were made of thin plastic that sagged under my feet as I walked, and the exterior windows were so heavily barred that it was difficult to see outside, but the tropical heat blasted in where panes were either broken or missing, giving the effect of a giant air-conditioner in reverse. It was easy to imagine that such a place, an underfinanced facility in a poor part of the world, might have trouble maintaining standards.
Sarmento is a big man, 6-foot-2 and slightly hunched, with a sad, doughy face all gathered up in worry. The crisis at the morgue had kept him from sleep, and he smiled wearily as he offered his hand. We took our seats by a table in his office, and he began to explain the crisis from two years earlier, with the fall of Flight 447.
“When we first found out, we were afraid,” he said. “We didn’t have space for 228 bodies. There were 33 nationalities on board, so we had to cooperate with other countries. We needed fingerprints, dental records, pictures of tattoos. We contacted Interpol right away, and they sent two people to work here and make the connection with other countries.”
Now Sarmento held up a finger with a look of irritation. “After one week,” he said, “the French government called and asked to send a representative to observe the autopsies.” Much of the forensic work took place at another site, but final examinations of the bodies were done in Recife. “When they got here,” Sarmento continued, “it was 20 specialists who wanted to do the autopsies by themselves. Only them. We couldn’t allow that. So I allowed one person from Interpol to be in the autopsy room and one person from the French government. Of course, this became a diplomatic issue.”
He reached for a large projector on the table and flipped it on with a hum. The far wall lighted up, and we began viewing images from the autopsies. “We took pictures of everything,” Sarmento said, scrolling through pictures of watches, necklaces, earrings and rings, still clinging to blue-green wrists and necks. “We were able to make all of the identifications.” As the images flashed by, he added: “All the autopsies were observed by the French and by Interpol. Not one country, not one family, complained about the identifications.”
After a while, Sarmento flipped off the projector and pushed away from the table in his chair. “Ninety percent of the passengers had fractures in the arms and legs,” he said. “Many of them also had trauma in the chest, in the abdomen, in the cranium. We didn’t find anybody burned.” He leaned forward in his seat and wrapped his arms around his knees. “They were like this,” he said, holding the crash position and looking into my eyes. Then he sat up quickly and held his hand flat above the table. “When they hit,” he said, slamming it down, “fractures. I believe the pilot tried to land in the water. This is consistent with the fractures. But when the bodies arrived, the lungs were already in a state of decomposition. We didn’t have conditions to see if anyone drowned.”
This hung in the air for a moment as I considered what he was suggesting.
“So it’s possible that some of them were still alive?” I asked.
Sarmento nodded. “Most died on impact,” he said. “Some could have survived.”
A few days later, in Paris, I stopped by the office of Alain Bouillard, the lead crash investigator for the B.E.A. After studying thousands of pieces of wreckage, Bouillard came to the same conclusion as Sarmento about the plane’s landing. Many of the items recovered, like meal carts, were found with their contents compressed from the bottom, and pieces of the plane’s underbelly were flattened as if struck from below. “There is a high probability that the aircraft landed in one piece,” Bouillard told me. “We are reasonably certain.”
“The medical examiner in Brazil didn’t see any signs of explosion,” I noted.
“No,” Bouillard said. “We are sure that there is no depressurization in flight, because all the masks were still in the box.”
“The medical examiner also said it’s possible that there could have been survivors,” I said. “Do you think so?”
Bouillard was silent. “I don’t know,” he said. “It’s impossible to say.”
Of course, some passengers may have survived the impact and then died quickly, but there is also a possibility that some lived longer. The surface water near Tasil Point can be as warm as 80 degrees in June, and according to hypothermic tables, a person can survive in those conditions for up to 12 hours before falling unconscious. The search plane finally arrived at Tasil Point 13 hours after the crash.
But what happened in those final hours may not be a mystery forever. On the Alucia this spring, as Woods Hole scientists scanned the first photos of Flight 447, they saw more than just landing gear, engines and wings. They also saw the bodies of at least 50 passengers sprawled across an abyssal plain at the base of the mountains. As they continued searching the area, they found a section of damaged fuselage not far away, large enough to contain more passengers. Members of the crew told me that a grim silence descended on the ship, and as word of the bodies reached around the world, an uneasy question began to rise. In the nearly frozen water, two miles down, the bodies would be extraordinarily fragile, but they might also be preserved more fully than those found floating on the surface two years earlier. Protected from light and nearly devoid of microorganisms, they might offer new answers not only about what happened to Flight 447 but also to the men and women aboard. The question was whether anyone really wanted to know.
ONE MORNING IN PARIS, I stopped by the apartment of Pérola Milman, a quantum physicist who lost her husband, Ivan, in the crash. Milman is a lean, athletic woman, originally from Brazil, with warm caramel skin and an aquiline nose. We sat facing each other in the living room while her children played nearby. Before the crash, Pérola and Ivan dreamed of moving into the city. After the crash, she finally made the move without him.
“I couldn’t stay at our house anymore,” she said “ I couldn’t do it. I had to move on. So I just left. I left all my furniture there. All my clothes. Everything. I had to do it.” She turned her head to watch the kids, then she said: “Children are amazing, you know? A psychologist told us after the accident that children do not have a sense of death until they are 6 or 7. And I radically contest this. When it happened, Jose was 4 years old. And of course, it’s a conversation I will never forget. I told him, ‘Listen, there was an accident with Daddy’s plane, and he’s not coming home.’ And he started crying as I never saw him cry before. He was saying, ‘But there are so many things I wanted to make for him.’ ”
Milman’s eyes were wet, but she went on. “I am a scientist. I know something concrete happened to the airplane. But I cannot prevent wanting the mystery. I don’t want them to bring up the bodies. I don’t want all that coming to the surface. I have this need to turn the page. It’s very strange to think that this place exists somewhere and my husband is there, in the same clothes he was wearing the last time I saw him, and his ring . . . and his necklace. . . .” Her voice trailed off, and she smiled.
The next day, I visited the apartment of John Clemes, whose brother, Brad, had flown to Rio on the airplane that would become Flight 447, which he reboarded when he discovered that he could not enter without a visa. After the accident, Clemes and Milman became friends, but the question of whether to bring up the dead hovered between them. As deeply as Milman felt the need to leave her husband below, Clemes felt an obligation to bring his brother home. “It’s horrible thinking they’re lost,” he told me. “There’s no body, no saying goodbye, it’s just . . . gone. For the first couple weeks, I just assumed they were going to find the plane. It wasn’t imaginable that they weren’t.”
Before the announcement from the Alucia last month, these differences could be set aside, distant and immaterial, but the discovery made them manifest. After less than 24 hours, the French minister of ecology and transportation, Nathalie Kosciusko-Morizet, pronounced on television, “These bodies will be raised and will be identified.” But whether that promise is possible to keep remains unknown. Now that the B.E.A. has recovered the black box cylinders, its crew will try to bring up at least one body, using the same claw-and-basket system. Forensic experts say this is dangerous and tricky. After two years, the bodies may be recoverable, but they will have a soft, fragile consistency that is likely to disintegrate in the robotic claw. Some experts say that the bodies should have been photographed exhaustively by 3D cameras before the recovery of the black boxes began. Already, the process of scouring the wreckage may have kicked up turmoil on the bottom.
Even if the bodies can be recovered, the question of whether they should be recovered remains difficult to answer. In the end, the decision is binary: either all the bodies will be recovered or they will all be left below. But this is a choice that could not have existed even a decade ago; it is one that could emerge only in the strange confluences of Flight 447. Because the plane was used so widely that its disappearance had to be explained. Because the only way to explain was by recovering the black box — an archaic device on its way to obsolescence but not there yet. Because the tools and technology to locate those boxes in the depths of the midocean ridge were available for the first time. And so, in order to solve the mechanical mystery, the closure for families would have to wait. They would have to wonder once again if their mothers and brothers and sisters were coming home. They would have to struggle with the possibility, each family alone, torn now among themselves, and sometimes even torn within.
STEPPING OFF an airplane in Brazil one afternoon, I noticed an e-mail marked urgent from Mary Miley, whose sister, Anne DeBaillon Harris, was on the plane with her husband, Mike, headed for two weeks in Paris. Anne and Mike were the only Americans onboard, and I had spoken with Miley perhaps a dozen times, sometimes for more than two hours. Something about Anne kept pulling me back. I would like to say that it was her eyes or her smile or some detail from the photos I saw of her, but Miley’s stories of her sister were not so nostalgic. She described Anne’s fight with cancer in her 20s, the ministroke on her left side, the migraines and bouts of fibromyalgia that would seize her for days on end. But I also heard from Anne’s friends, about how in Rio many of these problems seemed to fade, at least enough to give Anne, a girl from Lafayette, La., the taste of a life unlike anything she’d known, bustling through metropolitan markets, haggling with street vendors and learning to samba. I found myself exploring her neighborhood in Rio and driving into the hills west of town to stand at the edge of a terrace bar where she liked to sit with friends in the evening, gazing over the beach as waves crashed on the rocks below. And always, I would check back with Miley to share what I had seen and heard — and to ask more questions about Anne.
Now Miley had a question of her own. After the crash, Mike’s oil company closed their apartment and accounts and sent home all of their belongings — with one exception. Anne’s jewelry, she recently learned, was still at the oil company in Rio. She wondered if I would bring it home.
On my last morning in Brazil, I picked up three boxes of Anne’s jewelry from the oil company, then a few days later, I landed in Louisiana to deliver them to Miley. She met me at the door with a hug, and we sat down together at a table as she laid out Anne’s items between us — a turquoise ring, a wire bracelet, a necklace of tiny crimson beads — saying things like, “Oh, Anne, I don’t know about this one.” But when she opened the last box, she froze. The color drained from her face, and she reached inside the box with both hands, pulling out a string of pearls. She clutched them to her chest and squeezed her eyes. After a long silence, she looked up. “These are Mama’s pearls,” she said.
The day before, I called Miley to tell her about the discovery of Flight 447, and I knew she was still grappling with her feelings. For weeks, she had ricocheted between emotions. One day, she would say dismissively: “I don’t care if they find it. It’s not going to bring Anne back.” But the next, she would pepper me with questions about how the submarines really worked and if they really had a chance.
Now she sat quietly with the pearls. She rolled them between her fingers. “This may be the last thing I ever have of her,” she said. Her eyes were red and worn. “I do hope they can bring her home. When there was no reason to hope, I didn’t hope. But now that there’s hope, I hope.”