The Mystery of Air France Flight 447

One of aviation’s greatest mysteries had a stunning cause. Its legacy was equally troubling.

“Maintain flight level 350 (35,000 feet),” instructed air traffic control.

“Ok, will do,” Captain Marc Dubois replied. For now, no further communication was expected.

Fourteen minutes later, Air France flight 447 flew out of radar coverage, heading northeast over the Atlantic Ocean away from the coast of Brazil. Such instances are not uncommon on long overwater flights. The time was 1:49 a.m.

Two hours passed. On the other side of the Atlantic in Dakar, Senegal, air traffic controllers waited for flight 447 to show up on their screens as planned. It did not. They attempted to contact the aircraft.

Air France 447, this is Dakar. Do you copy?”

There was no response.

11:10 a.m. June 1, 2009. Air France flight 447 was scheduled to arrive at Charles de Gaulle Airport in Paris. Friends and relatives of many of those on board gathered eagerly at Terminal 2E to welcome their loved ones arriving off the long flight from Rio. The aircraft failed to turn up. Minutes turned into hours. Eventually, the news broke.

Flight 447 had vanished.

Search and rescue crews scrambled to the South Atlantic in search of the missing plane, based on automated position reports sent from the aircraft at ten-minute intervals. The search area was vast — the last position report showed that the aircraft was almost midway between the continents of Africa and South America, and there was no way of telling which direction it took thereafter. For days, the search turned up nothing.

Five days later, on June 6, debris was spotted floating on the surface of the Atlantic, as were some passenger baggage and two bodies. The next day, the tail of the aircraft was found, with its unmistakable blue, white and red stripes still visible, erasing any lingering doubt that the wreckage found was indeed from the missing plane. It was soon clear that Air France flight 447 had crashed into the Atlantic Ocean, killing all 228 passengers and crew on board. It was the deadliest accident for both the airline and the Airbus A330, as well as the worst aviation accident in the world since 2001.

For investigators from the French Bureau of Enquiry and Analysis for Civil Aviation Safety (BEA), though, the recovered pieces from the surface of the ocean were woefully inadequate for their needs. Modern airliners are equipped with two flight recorders — the cockpit voice recorder and flight data recorder, through which investigators are able to find out what happened onboard the aircraft in the moments leading up to the accident. However, being dense, heavy objects, the recorders had almost certainly sunk to the bottom of the Atlantic Ocean along with the main wreckage, in an area reaching up to four kilometres in depth. The seabed in this area was uncharted territory, and recovering the main wreckage of the A330 was going to be an uphill task.

Fortunately, investigators were not completely out of leads. The aircraft had been sending periodic position reports to Air France headquarters every ten minutes, as well as maintenance messages when relevant, through a system known as the Aircraft Communications Addressing and Reporting System, or ACARS for short. The ACARS messages contained information on the conditions of the various systems on board, as well as any abnormalities encountered, to prepare the maintenance crew on the ground before the arrival of the aircraft. Shortly after the last routine position report, 24 messages were transmitted over a duration of fewer than five minutes, between 2:10 a.m. and 2:15 a.m. Investigators began to scrutinise these messages carefully. Soon, a clue was found: one of the messages indicated a fault with the aircraft’s pitot tubes. The pitot tubes are small sensors protruding from the front of the aircraft pointing forward to measure airspeed and are crucial in the operation of a modern aircraft. Without them, the speed of the aircraft cannot be determined accurately. This could lead the aircraft to either overspeed or stall, resulting in loss of structural integrity or lift respectively.

Why would such a fault occur? Investigators honed in on a potential culprit: flight 447 had crashed in an area known as the Intertropical Convergence Zone (ITCZ), the region around the equator where the prevailing winds from both hemispheres converge. Violent thunderstorms reaching up to high altitudes are common here, especially during the monsoon periods.

Studying meteorological reports in the region from the night of the crash, investigators discovered that flight 447 had indeed been flying through one. The conditions of the storm were ripe for the formation of ice crystals, which could accumulate within and block the pitot tubes, causing them to provide inaccurate data. In fact, in the year prior to the crash, nine such incidents had been reported on the Airbus A330 and A340 fleet of Air France. The airline was in the process of replacing the pitot tubes on their A330s and A340s with a newer model that was not susceptible to this problem, but those on the aircraft involved in the accident had yet to be replaced. Investigators were convinced that flight 447 encountered icing conditions while flying through the storm, causing its pitot tubes to be blocked and fail.

But this alone could not explain the crash. In the nine previous instances of this problem, the blocked pitot tubes cleared after less than a minute, and the aircraft involved landed safely. Pilots at Air France, including those on flight 447, had been trained on how to handle the problem — simply maintain a stable attitude and sufficient engine power while waiting for the pitot tubes to unclog. As long as this was achieved, airspeed would not fluctuate to a dangerous extent during the period of inaccurate readings. How, then, could a seemingly minor and temporary problem cause a modern airliner like the A330 to fall out of the sky? Investigators desperately needed more evidence to find out.

Meanwhile, out at sea, salvage teams worked around the clock to locate the sunken wreckage of Air France flight 447. The flight recorders on a modern airliner are each fitted with an Underwater Locator Beacon (ULB). The ULBs emit ultrasonic pulses every second when immersed in water, which can be detected by equipment onboard ships from the surface of the ocean. However, their batteries last only 30 days. If the wreckage was not located by then, the chances of finding out what happened over the South Atlantic Ocean on June 1, 2009, would be greatly diminished.

The clock was ticking. The French Navy dispatched a nuclear submarine to assist in the search, while the French Research Institute for Exploitation of the Sea sent the Nautile, the submersible which had examined the wreck of the RMS Titanic in 1987. Despite the best efforts of those participating in the search, it seemed like the aircraft had simply vanished into thin air. The 30-day mark came and went. With the ULBs no longer functioning, multiple searches ensued to comb the seabed for the remains of the A330. No further traces of the aircraft were found, and the search ended on August 20, 2009.

Though only a few small fragments of the aircraft had been found, the recovered pieces did provide a few clues. The galley fittings, luggage racks and vertical stabiliser showed signs of severe vertical compression. This indicated that flight 447 hit the surface of the ocean at a high vertical speed in an approximately level orientation. Several passenger oxygen canisters had also been recovered, and an examination of them showed that they were in the closed position. This meant that the oxygen masks had not been deployed, and there was thus no depressurisation of the cabin in flight. The fuselage of the aircraft was intact at the point of impact.

Autopsies of the recovered bodies told a similar story. The victims had sustained severe spinal compressions along with fractures of the pelvis, indicating that they had been subjected to a large upward vertical acceleration while seated. Equally important was what was not found — their lungs had not been damaged, which would point towards a sudden decompression; nor were their extremities badly dislocated or fractured, which would indicate that they had been ejected from the aircraft at high altitude. It was clear that the passengers and crew onboard flight 447 had died from impact forces when the aircraft slammed into the ocean.

It was thus established that flight 447 was falling vertically in an upright orientation when it crashed. There was only one possible explanation for the aircraft to make such a manoeuvre — it was stalling.

There were still many gaps in the narrative. Why did the aircraft stall? Why were the pilots unable to recover from the stall? And what did the iced-over pitot tubes have to do with this? With the flight recorders still missing, investigators were down to their last available leads — the history of the aircraft and the backgrounds of the pilots at the controls of the flight.

F-GZCP, the aircraft involved in the crash, was an Airbus A330–200 powered by two General Electric CF6–80E1A3 turbofan engines. Just four years old at the time, it was the newest A330 in the Air France fleet, delivered to the airline on April 18, 2005. The aircraft had sustained minor damage when it was involved in a ground collision with an Airbus A321 at Charles de Gaulle Airport in 2006, but it was repaired and returned to service shortly after. In the week before the crash, F-GZCP had made multiple flights to Bangalore and Cairo, as well one to Philadelphia and back, without any issues. It then set off on a round trip to Rio de Janeiro in the morning of May 31, departing as flight 444 and scheduled to return the next day as flight 447.

While the A330 is usually flown by a crew of two pilots, there were three onboard flight 447, comprising a captain and two first officers. Due to the long flight time from Rio de Janeiro to Paris, scheduled at 11 hours and 15 minutes, the three pilots would take turns to fly the aircraft, with one pilot on a rest break at any point.

Captain Marc Dubois was 58 years old and had been flying for over 30 years, beginning his career in general aviation before eventually becoming an airline pilot in 1988. He had amassed a total of 10,988 hours in the air and was a very experienced pilot by any measure. However, he was still rather new to the A330, having only flown it since 2007, but had flown the A320, a smaller aircraft with similar cockpit characteristics, for almost ten years before switching to the A330.

First Officer David Robert was 37 years old and had clocked 6,547 flying hours since joining Air France in 1998. With 4,479 of these hours on the A330 over seven years, he had the most experience on the A330 among the three pilots aboard the flight that night. Having flown to South America 39 times over his career with the airline, he would have been very familiar with the route that night. Robert had also in fact been at the controls when the crew landed in Rio three days before their departure back to Paris on flight 447.

First Officer Pierre-Cédric Bonin, 32, was the most junior of the pilots on flight 447. He had joined the airline less than six years ago, in late 2003, and had accumulated 2,936 flight hours. Bonin had only been flying the A330 for about a year, but like Dubois, he had been flying the similar A320 before that. Nonetheless, he was fully qualified to fly the aircraft that night, as were the other two pilots on the flight.

Nothing in the records of the plane or the pilots raised any red flags. Investigators were back to square one.

On April 2, 2010, another operation was launched, involving two ships, a sonar array from the US Navy, and three autonomous underwater vehicles (AUVs) designed and operated by the organisation which first discovered the Titanic in 1985. Oceanographers from all around the globe drew up a large area based on the last known position of the aircraft, the location of the floating debris found, and the currents in the Atlantic that would have moved the wreckage over the months. The extensive search combed through this area, as well as a smaller area based on a re-analysis of sonar recordings made during the first search. Despite combing through a vast 6,300 square kilometres of the ocean floor, the search returned nothing, and ended on May 24. The investigation into the crash ground to a halt owing to a lack of evidence. It seemed that whatever brought down flight 447 on June 1, 2009 would forever remain a mystery.

In March 2011, yet another search operation was planned. It had been almost two years since Air France flight 447 disappeared over the Atlantic Ocean on a routine flight from Rio de Janeiro to Paris, and locating the main wreckage had morphed into the longest and most expensive undersea search to date. One final search was authorised — if it failed to return anything, the wreckage of flight 447 would be lost forever. The search was launched on March 25, again making use of AUVs. A week later, on April 3, one of the AUVs captured a sonar image of what seemed like something large lying on the seabed.

A remotely operated submersible was dispatched to the location to capture high-resolution digital images. Investigators could not believe their eyes as they took in the images of what was clearly the wreckage of a large airliner.

By early May, both the cockpit voice recorder and flight data recorder were recovered. Amazingly, they were in excellent condition despite spending two years subject to the crushing hydrostatic pressure four kilometres below the surface of the Atlantic. French investigators were finally able to access the crucial data stored in their memory cards.

The cockpit voice recorder records two hours of audio recordings from the aircraft’s cockpit, allowing investigators to listen to the conversations between the pilots, any problems they may have discussed, and as any alarms going off in the cockpit; while the flight data recorder records various flight parameters such as the airspeed, altitude and attitude of the aircraft, as well as the positions of the controls. Together, they paint a vivid picture of the flight. As investigators examined the data from both flight recorders, the full story of what really happened on board the aircraft emerged.

As Air France flight 447 pushed back from the gate at Galeão International Airport in Rio de Janeiro on May 31, 2009, the three pilots gathered in the cockpit for the departure. Due to the long flight time, they would take turns to rest while the other two flew the plane. Captain Marc Dubois occupied the left seat, with First Officer Pierre-Cédric Bonin to his right, while First Officer David Robert sat in the jumpseat behind the two. Robert would take the first break, followed by Dubois and finally Bonin. The flight was filled nearly to capacity, with just four empty seats throughout the vast passenger cabin. After a short four-minute taxi, the A330 took off at 7:29 p.m. local time for its twelve-hour flight to Paris, with 216 passengers and 12 crew aboard. The aircraft climbed out of Rio, and as it passed through 20,000 feet in altitude, Robert left the cockpit for his scheduled rest break.

(Note: All further times specified are in Universal Coordinated Time.)

For the next two hours, nothing was untoward as flight 447 headed northeast on its planned route, skirting the coastline of Brazil. The aircraft was now cruising at 35,000 feet. As they discussed the flight’s route, however, Dubois noticed something was up with Bonin.

“Try maybe to sleep twenty minutes when (Robert) comes back, or before, if you want,” Dubois offered. Though he was scheduled to take the next break, he sensed that Bonin might need the rest more than he did.

“Yeah ok, that’s kind. For the moment, I don’t feel like it, but if I do feel like it yeah,” Bonin replied.

“It’ll be a lot for you,” Dubois added, but Bonin declined the captain’s offer. Flight 447 headed out over the Atlantic Ocean on its planned route, leaving the South American continent behind.

An hour later, as the A330 flew out of Brazilian radar coverage and into the communications dead zone over the mid-Atlantic, Bonin started to get worried. He noticed a storm was showing up on the aircraft’s weather radar in their path.

“So, we’ve got a thing straight ahead,” Bonin said.

“Yes, I saw that,” Dubois replied. He was not too concerned — it did not appear to be a particularly severe one, and in his multiple decades of experience piloting aircraft, he had probably flown through worse conditions before.

Normally, pilots either try to fly above a storm or navigate around it. However, the Airbus A330 was still laden with fuel for the long transatlantic flight and was too heavy to climb any higher. Besides, the storm seemed to be a relatively minor one. The pilots decided that they would fly through the storm.

Minutes later, as flight 447 entered the outer edges of the storm, Dubois turned on the aircraft’s exterior lights to make out its severity. “It’s snowing,” he remarked.

Bonin dimmed the cockpit lights to get a better view outside. “It looks like we’re entering the cloud cover,” he told the captain.

The aircraft began to experience moderate turbulence. Dubois and Bonin decided that they would try to climb above the storm as soon as they had burned enough fuel to do so, which they expected would take at least another hour. It was now 1:50 a.m. on June 1. First Officer Robert would soon return from his rest break. Captain Dubois, the most senior member of the cockpit crew, would take the next break, as the middle of the flight was typically the most uneventful. But now, their windshields were illuminated with St. Elmo’s fire, plasma created by a corona discharge often observed in severe thunderstorms. “All we needed was Mr St. Elmo,” Dubois quipped. “It’s going to be turbulent for my rest.”

At 1:56 a.m., Dubois pressed a button to trigger a chime in the crew rest area, informing Robert that his break was over. Robert responded with a knock on the partition separating the rest area from the cockpit. Three minutes later, he entered the cockpit.

“Did you sleep?” Bonin queried Robert.

“So-so,” Robert replied.

Dubois was concerned. “You didn’t sleep?” he asked.

“I was dozing,” Robert clarified.

“Well then,” Dubois announced to the two first officers as he unfastened his seat belt, “I’m out of here.”

After briefing Robert on the radio frequencies to contact air traffic control in Senegal later on in the flight, Dubois left the cockpit for his scheduled rest break at 2:01 a.m. The aircraft was now in an area of calm air, and it was shaping out to be yet another routine flight. He had no idea that in less than fifteen minutes, everyone on board would be dead.

Back in the cockpit, Robert settled into the left seat for the transatlantic journey. The captain had moved the seat backwards as he left, and Robert saw no reason to move it forward. He lay comfortably back in his seat and propped his feet up on the A330’s footrests, expecting a typical relaxed cruise. It was an awkward position to operate the controls from, but pilots are rarely required to fly the aircraft manually during cruise.

The weather radar showed more stormy weather ahead.

“In two minutes, we ought to be in an area where things will start moving about a bit. You’ll have to watch out there,” Bonin told a flight attendant through the cabin interphone.

“Alright, are we going to sit down?” the flight attendant asked him.

“I think that might be a good idea,” Bonin replied. “I’ll call you when we’re out of it.”

Meanwhile, Robert toggled the weather radar display and was alarmed by what he saw. The storm was much larger and more severe than Dubois and Bonin had previously thought. “Do you maybe want to go to the left a bit?” he asked Bonin, who had been designated as the pilot flying for this segment. Bonin turned the aircraft twelve degrees to the left, but this did little to avoid the worst of the storm — they were already too far in.

As flight 447 flew deeper into the storm, Bonin noticed a strange smell in the cockpit.

“You did something to the A/C?” he queried Robert.

“I didn’t touch it,” Robert answered.

“What’s that smell?” Bonin asked.

“It’s ozone, that’s why,” Robert replied. He knew this was a somewhat common occurrence when flying through storms, particularly those in the ITCZ.

Bonin was still ill at ease. “You can feel already that it’s a lot hotter,” he observed.

Less than two minutes later, the A330 flew into a column of ice crystals, which pounded the fuselage audibly. Bonin slowed the aircraft slightly to ride out the coming turbulence.

Twenty seconds later, at 2:10 a.m., the pitot tubes became clogged with ice. As the airspeed indicators in the cockpit began to show different readings, the flight computer was no longer sure which was correct.

The A330 is a fly-by-wire aircraft, meaning that pilot inputs are sent to a computer via electronic signals as a description of how the aircraft should behave. The computer then processes the input to determine how the control surfaces such as the ailerons, elevators and rudder should be moved to achieve that outcome. Under normal circumstances, it is also able to override the inputs of the pilots if it deems that fulfilling those commands would endanger the aircraft. However, due to the loss of valid airspeed data, the autopilot switched to a state known as “alternate law”, a mode offering degraded flight protections. In this mode, the autopilot would not guard against overspeed or stall situations. A warning blared in the cockpit informing the pilots of the switch.

“I have the controls,” Bonin declared.

“Alright,” Robert acknowledged.

Flight 447 was now in manual control. All Bonin had to do was keep the aircraft flying straight and level while the pitot tubes cleared, a process which would typically take about a minute. After that, with valid airspeed readings back, the autopilot could be re-engaged and the aircraft could carry on normally to Paris. However, he was unnerved by how quickly things seemed to be changing, and reacted irrationally. Instead of holding the aircraft level, Bonin pulled back on his sidestick, raising the nose and putting the A330 into a steep climb from 35,000 feet, a manoeuver that generated a vertical acceleration of 1.6G. “STALL! STALL!” a synthetic voice screeched.

Robert was caught off-guard. “What’s that?” he asked.

“We haven’t got a good display of speed,” Bonin answered. He seemed to be convinced that the stall warning was inaccurate due to the loss of reliable airspeed data.

Scanning his instruments, Robert now noticed the aircraft was climbing. “Go back down,” he instructed Bonin. He cross-checked his altimeter with Bonin’s, as well as against the standby instrument, to ensure that his readings were accurate. “According to all three, you’re going up. So, go back down.”

“Okay,” Bonin replied. But inexplicably, he continued to pull back on his sidestick.

Seeing as the situation was getting dire, Robert pushed the button triggering the chime in the crew rest area to get the captain back to the cockpit. “Damn it, where is he?”

Like a car going uphill, flight 447 slowed as it climbed, decreasing the amount of lift generated by the wings. The airframe began to shudder as the wings struggled to keep the aircraft aloft. Bonin applied maximum thrust on the engines in an attempt to gain speed, but at such a high altitude, the air outside was too thin for this to have any meaningful effect.

Just over one minute since the pitot tubes were clogged, the last of the icing cleared, as had been the case during the nine previous occurrences of this issue over the past year. All airspeed indications in the cockpit were now accurate, and showed an airspeed of just 183 knots, well below the normal 260 knots at their altitude. Bonin continued to pull back on his sidestick, increasing the already high angle of attack of the wings. This continued up to and beyond a point where the wings effectively acted like walls into the incoming flow of air and were no longer able to provide lift. Seconds later, the aircraft reached a maximum altitude of 37,924 feet. The A330 entered an aerodynamic stall, falling like a rock towards the ocean below.

“But we’ve got the engines! What’s happening?” Robert was now just as confused.

The aircraft began to roll to the right. “Do you understand what’s happening or not?” Robert asked.

“I don’t have control of the airplane any more now. I don’t have control of the airplane at all,” Bonin replied.

“Controls to the left,” Robert announced, indicating that he would be taking over control of the aircraft from Bonin. He held his sidestick fully to the left in an attempt to correct the right bank. But he did nothing to correct the high nose-up attitude of the aircraft, which had been the source of their problems the whole time. Instead of releasing the controls to Robert, however, Bonin continued to pull back on his sidestick.

98 seconds after the initial blockage of the pitot tubes, Captain Dubois returned to the cockpit. It had been only ten minutes since he first left the cockpit for his break. “What are you doing?” he asked the two first officers.

“What’s happening? I don’t know, I don’t know what’s happening,” Robert told the captain.

“We’re losing control of the aeroplane,” Bonin said.

“We don’t understand anything. We’ve tried everything,” Robert added.

Dubois was perplexed. The nose of the aircraft was pitched up, the engines were at maximum power, and yet they were falling from the sky at a rate of 15,000 feet per minute. Because the sidesticks on the A330 are positioned to the side of each pilot’s seat away from the centre of the cockpit, no one could see that Bonin’s actions were causing the aircraft to stall, especially at night in a dark cockpit. Under such an environment, the only reasonable way the others could know that Bonin had been pulling the aircraft’s nose up was if he told them so — and so far, he had not. By now, the indicated airspeed was so low that the flight computers assumed them to be erroneous, and stopped setting off continuous stall warnings in the cockpit.

“I don’t have vertical speed information,” Bonin observed. The aircraft’s vertical speed was far outside of the maximum display range of the indicator, causing it to disappear from the screen.

“We have no more valid displays,” Robert added.

“I have the impression that we have some crazy speed. What do you think?” For some reason, despite all the indications that the aircraft was stalling, Bonin seemed to think that they were flying too fast, and deployed the aircraft's spoilers.

“No, no. Above all, don’t extend the…” Robert cautioned.

“No? Okay,” Bonin replied, and retracted the spoilers.

“Don’t extend,” Robert confirmed.

“So we’re still going down,” Bonin pointed out.

“What do you think about it? What do you think we need to do?” Robert asked Dubois.

“I don’t know,” Dubois answered exasperatedly, “it’s going down!”

Under the immense stress and possible fatigue, all three pilots were now utterly confused as to their aircraft’s predicament.

“You’re climbing,” Robert pointed out to Bonin. “You’re going down,” he corrected himself a second later.

“Am I going down?” Bonin asked.

“Go down,” Robert said.

“No, you’re climbing,” Dubois chimed in.

“I’m climbing. Okay, so we’re going down,” Bonin appeared to contradict himself.

“It’s impossible,” Dubois said, not appearing to believe the situation in front of him.

“I’m going down, no?” Bonin queried again.

“You’re going down, yes,” Robert confirmed.

All of a sudden, the A330 rolled 41 degrees to the right.

“Hey, you… You’re in…” Standing in the area behind and in between the two pilot seats, Dubois struggled to keep his balance.

“Get the wings horizontal,” he directed.

“That’s what I’m trying to do,” Bonin informed the captain as he attempted to level the wings of the aircraft.

“Get the wings horizontal,” Dubois repeated.

“I’m at the limit with the roll,” Bonin replied. Over the next few seconds, flight 447 rolled slowly back through the wings-level position and settled in a shallow left bank.

The aircraft was now falling through 10,000 feet. The descent rate had slowed slightly due to the higher air density, but the aircraft was still stalled. At the current rate, they would hit the water below in about a minute.

“I have the controls,” Robert declared once again. He pushed his sidestick forward to recover from the stall, but Bonin was still holding his sidestick back. Their inputs cancelled each other out, and there was no change to the pitch of the aircraft. “DUAL INPUT,” a synthetic voice announced.

“How come we’re continuing to go right down now?” Bonin asked, still oblivious to the impact of his own actions.

“Try to find what you can do with your controls up there,” Robert suggested to Dubois. “The primaries and so on,” he continued, referring to the A330’s primary flight computers. He seemed to think the problem was with the electronic systems on board the aircraft.

“It won’t do anything,” Dubois advised Robert, but proceeded to reset the computers anyway. As he expected, nothing happened.

“Nine thousand feet,” Bonin called the aircraft’s altitude out to the crew. The Atlantic Ocean was mere seconds away.

“Climb, climb, climb, climb!” Robert muttered to himself.

“But I’ve been at maximum nose up for a while!” Bonin replied, finally revealing this crucial piece of information to the rest of the crew.

At last, Dubois understood what had been going on: Bonin was causing the stall by pulling the nose up.

“No! No! No! Don’t climb!” Dubois exclaimed.

“So, go down,” Robert instructed, and pushed his sidestick forward again. But Bonin was still pulling back. “DUAL INPUT,” the synthetic voice chimed.

“Give me the controls!” Robert urged Bonin.

“Go ahead, you have the controls,” Bonin replied. Despite saying this, he continued to pull back on his sidestick.

“DUAL INPUT,” the synthetic voice announced again.

“Watch out, you’re pitching up there,” Dubois pointed out.

“Well we need to, we are at four thousand feet!” Bonin insisted.

In any event, it was already too late. There was not enough altitude remaining for the A330 to recover. Nine seconds later, the aircraft’s sensors detected the fast-approaching ocean surface. “SINK RATE. PULL UP,” sounded the ground proximity warning system.

“We’re going to crash,” Bonin muttered in despair. “This can’t be true! But what’s happening?”

“Ten degrees pitch…” Dubois began.

Before he could finish his sentence, it was all over. At 2:14:28 UTC, Air France flight 447 slammed into the Atlantic Ocean, instantly killing all 228 passengers and crew, and sparking a search and recovery effort that would go on for the next two years.

The BEA released their final report into the crash of flight 447 on July 5, 2012, more than three years after the accident. Recommendations were made to strengthen and improve pilot training in situations involving degraded automation and flight protections and reinforce their knowledge of flight mechanics. Furthermore, given the unique nature of the accident and the prolonged search operation following it, recommendations were also made to improve the tracking of flights over large water bodies and other locations where radar coverage is spotty. These recommendations ranged from requiring that airliners be made to regularly transmit basic flight parameters such as position, altitude, speed, and heading via satellite, to extending the battery life of the ULBs on the flight recorders to 90 days, and installing an additional ULB as a backup should the existing ones fail.

Though shocking at the time, the crash of Air France flight 447 soon faded into the annals of history shortly after the final report was published. Then, on March 8, 2014, Malaysia Airlines flight 370 vanished over the South China Sea en route from Kuala Lumpur to Beijing.

As more details emerged over the following days, it was painfully clear that the recommendations following the crash of flight 447 had not been followed. The tracking system on flight 370 was just as rudimentary as the one on flight 447, allowing the aircraft to fly undetected for hours away from its last known position. The lifespan of the ULBs was still 30 days, and much of that time was spent searching the wrong location, meaning that when salvage equipment finally reached the presumed resting place of the wreckage, the batteries on the ULBs had run flat. Days, months, and then years passed, without a trace of the Boeing 777. The search for the wreckage of flight 370 would eventually exceed that of flight 447 in both duration and cost, with the first pieces of the aircraft being found more than a year later.

Less than five years separated the crashes of flights 447 and 370, yet the events following them were remarkably similar. Did the 228 people who lost their lives on board Air France flight 447 do so in vain? While we may never know for sure, it does seem that the search for the wreckage of flight 370 has been needlessly long and expensive, especially when compared to the recommendations made by the BEA in the final report of flight 447. If only those recommendations had been acted upon promptly, we may very well have already known the full story behind flight 370.