An Airbus A330 is on a routine overnight flight from Rio de Janeiro to Paris. As it cruises over the vast darkness of the Atlantic Ocean at 35,000 feet, inside the cabin is quiet and peaceful. Many passengers are probably sleeping.

All of a sudden, the autopilot disconnects and the aircraft begins rolling to the right.

Within seconds, conflicting alarms start blaring in the cockpit. Airspeed indications fluctuate wildly. “I have the controls,” calls the co-pilot.

Less than four and a half minutes later, the aircraft falls nose-up into the ocean at a vertical speed exceeding 10,000 feet per minute.

Tragically, all 228 souls are lost.

For two agonizing years, the truth remained locked inside flight recorders almost 13,000 feet below the Atlantic’s surface. Grieving families bore the terrible pain of not knowing exactly what happened in their loved ones’ final moments.

When investigators finally recovered the black boxes, the devastating sequence of events they discovered would raise an unsettling question: who was really flying the plane in those final few minutes?

This is the story of Air France Flight 447.

The media often refers to Air France 447 as “the Titanic of the skies.” It’s an evocative metaphor comparing two advanced technological marvels from very different times, both of which met tragic ends that shocked the world.

For almost two years, a haunting question remained unanswered: how could one of the world’s most sophisticated passenger aircraft simply ‘fall from the sky’?

This is a story of how human psychology can unexpectedly collide with advanced technology when met with a perfect storm of adverse conditions.

On May 31st, 2009, an Airbus A330-203 prepared for departure from Rio de Janeiro Galeão airport. The aircraft, with registration F-GZCP (Foxtrot-Golf-Zulu-Charlie-Papa), had been in service for just four years.

Routine maintenance had been performed according to schedule, and the aircraft had no significant technical issues beyond a minor problem with one of the radio panels.

The flight crew consisted of three experienced pilots.

Captain Marc Dubois, aged 58, had logged nearly 11,000 flight hours, including more than 1,700 on the A330.

First Officer Pierre-Cédric Bonin, 32, designated as Pilot Flying for this leg, had accumulated almost 3,000 total flight hours, with 807 on type. His wife Isabelle, a physics teacher, was among the passengers aboard for this trip.

Air France’s procedures for long-haul flights required augmented crew to allow for rest periods during the 11-hour journey. The flight was scheduled to depart Rio de Janeiro Galeão airport at 10:00 p.m. local time, arriving in Charles de Gaulle, Paris the following morning.

This overnight flight meant the crew would be operating through what aviation physiologists call the “window of circadian low” – typically between 2:00-6:00 a.m. body clock time. During this period, the human brain experiences significant decreases in alertness, decision-making ability, and situational awareness, even in well-rested individuals. This physiological challenge is one key reason Air France, like most carriers, schedules augmented crews for long-haul flights.

The relief pilot, First Officer David Robert, aged 37, had over 6,500 flight hours in total. With almost 4,500 hours on the A330, he actually had more ‘on type’ experience than the captain and other First Officer combined. Robert would be designated as Pilot Not Flying or Pilot Monitoring. He was primarily in a management position at Air France’s operations center, flying occasionally to maintain his credentials.

Loading was completed with 216 passengers and 12 crew members including the 3 pilots. The passengers represented a global community from 32 different nations. French citizens constituted the largest group (61), followed closely by Brazilians (58) and Germans (26). Other nationalities included Chinese and Italian (9 each), Swiss (6), British (5), and two Americans.

The aircraft carried 70.4 tons of fuel at takeoff, more than sufficient for the planned route with adequate reserves for contingencies.

As the crew conducted pre-flight preparations, they were aware that they would be crossing the inter-tropical convergence zone – an area where trade winds from the northern and southern hemispheres meet, often creating bands of significant convective activity including thunderstorms. This is part and parcel for flights along this route.

Weather forecasts showed nothing exceptional for the time of year. SIGMET advisories indicated possible convective activity, but multiple daily flights navigated this same airspace without any issues. Satellite imagery later revealed a large cluster of storm cells along the route – powerful cumulonimbi with tops reaching above the aircraft’s cruising altitude – but these conditions were within normal operational parameters.

Following a delay of almost half an hour, the flight departed at 10:29 p.m. local time, climbing to its cruise altitude of 35,000 feet. Captain Dubois and First Officer Bonin managed the initial phase of the journey, while First Officer Robert rested in the crew rest compartment beside the cockpit.

All good so far. The night ahead looked like any routine oceanic crossing.

There’s no radar coverage in most oceanic regions, meaning air traffic controllers can’t “see” aircraft on their screens. Instead, planes follow predetermined tracks and report their positions at specified points.

At 01:35, Air France 447 reaches the INTOL waypoint, marking the transition from Brazilian radar coverage to oceanic airspace controlled by the ATLANTICO center. The crew makes radio contact, reporting their position and providing estimates for upcoming waypoints. They attempt to establish what’s known as an ADS-C data connection with the DAKAR Oceanic center but are unsuccessful. This satellite-based position reporting system would have automatically transmitted the aircraft’s position to air traffic control.

“So we’ve got a thing straight ahead,” Bonin remarks, referring to some weather in their path, adjusting his navigation display. Captain Dubois acknowledges, and the two discuss the outside air temperature, which is running higher than forecast. This means, in practice, that the aircraft would be unable to climb up to Flight Level 370, which might have provided a smoother ride above the developing weather. Aircraft have a ceiling of how high they can safely fly. This is determined by how thick or thin the air is, and because warmer air is less dense with the air molecules further apart, this essentially translates into less lift available for the aircraft’s wings.

Ten minutes later, at 01:45, the aircraft entered a zone of light turbulence approaching the SALPU waypoint. The crew dims the cockpit lighting and turns on the external lights to improve visibility. “We’re entering the cloud layer,” Bonin notes, adding that “it would have been good to be able to climb.” The turbulence continues for several minutes before subsiding.

The higher-than-expected temperature has pushed their maximum recommended altitude – known as “REC MAX” – to Flight Level 375, or 37,500 feet, just out of practical reach.

“It’s going to be turbulent when I go for my rest,” Captain Dubois said, his words heavy with an irony only fate could comprehend. He wakes First Officer Robert, who has been sleeping in the crew rest compartment.

“He’s going to take my place,” Dubois tells Bonin, referring to Robert who has just entered the cockpit and will now occupy the left seat. This was the beginning of a subtle power dynamic: the less experienced pilot (Bonin) was designated as Pilot Flying, while the more experienced and management-level co-pilot (Robert) took the monitoring role otherwise known as Pilot Not Flying.

It’s approximately 02:00 am. With Captain Dubois about to leave the flight deck, First Officer Bonin briefs Robert. “Well the little bit of turbulence that you just saw we should find the same ahead. We’re in the cloud layer unfortunately we can’t climb much for the moment because the temperature is falling more slowly than forecast. So what we have is some REC MAX a little too low to get to three seven” explains Bonin.

So in other words, the First Officer was explaining that they expect more turbulence ahead and the warm thinner air is making it unsafe to climb to 37,000 feet to escape the weather.

He also mentions the failed logon with DAKAR. Remember, this meant the aircraft couldn’t establish the satellite connection that would automatically report its position to air traffic controllers, essentially making the plane even less visible over the open ocean. The captain departs for his rest period, leaving the two co-pilots in control.

Bonin had expressed concerns even before the Captain left. He had mentioned the turbulence several times and suggested climbing to flight level 360 to avoid the cloud layer, saying “it would have been good to be able to climb.” We’ve no way of knowing exactly what was going on inside Bonin’s head, but it is reasonable to assume that his preoccupation with the weather situation revealed a level of anxiety that may have been building in his mind.

The handover between Captain Dubois and Relief Pilot First Officer Robert is a critical moment in this flight. Good Crew Resource Management (CRM) requires crystal clear communication – and it seems that the captain never explicitly said who would be in command in his absence. Also, the captain didn’t leave any specific instructions on how they should cross The Intertropical Convergence Zone.

The aircraft approaches the ORARO waypoint, maintaining Flight Level 350, or 35,000 feet, at a cruising speed of Mach 0.82. The two co-pilots discuss the temperature and “REC MAX” again, noting its gradual increase. The turbulence intensifies slightly.

At 02:06, Bonin lets the cabin crew know that they’re going to be shortly entering some turbulence.

Approximately two minutes later, at 02:08, Robert suggests, “Go to the left a bit…” The heading mode is activated, and they change course by about 12 degrees, deviating from their planned route to avoid weather that was detected on their radar.

Inside the cockpit, the smell of ozone became noticeable. According to the investigation’s final report for this flight, Bonin didn’t seem to recognize this environmental cue as they entered this meteorologically active area. “What’s that smell, now?” he asked. “It’s… it’s ozone,” replied the more experienced Robert from the left seat. Ozone is a distinctive electric scent often present during thunderstorms, a natural warning sign that they were entering an active weather system.

Robert adjusts the weather radar’s gain up to maximum. The crew decides to reduce speed slightly to about Mach 0.8, and they activate the engine anti-icing systems.

These routine decisions – subtle course changes, minor speed adjustments, system activations – are all standard practice carried out by a professional crew when navigating through areas of convective weather.

By now, the sound of ice crystals impacting the aircraft could be heard. Another telltale sign of being in the middle of an active thunderstorm.

Little did the pilots know, what lay ahead of them wasn’t just a weather system. They were unknowingly approaching turbulence of a very different nature: the complex interplay of weather conditions, human psychology, and automated technology that was designed to keep them safe, but would soon become part of the problem.

Suddenly, at exactly 02:10:06 am, the autopilot disconnects with a distinctive cavalry charge alarm. The aircraft begins rolling to the right. “I have the controls,” calls First Officer Bonin from the right seat. Almost immediately, a stall warning – a synthetic voice calling “Stall, stall” – triggers briefly, twice in succession. Within ten seconds, Bonin’s nose-up input increases the aircraft’s pitch attitude from 2 degrees to 11 degrees. The aircraft begins to climb rapidly, despite the engine thrust being insufficient for such a maneuver at this altitude.

The consequence is immediate and severe: the speed indications on the left primary flight display (PFD) plummet from approximately 275 knots to 60 knots in seconds. Shortly afterward, the same occurs on the standby airspeed indicator.

Robert asks, “What is that?” in response to the various warnings.

Bonin replies, “We haven’t got a good…” followed by “We haven’t got a good display… of speed” .

As he was saying this, First Officer Robert called out “We’ve lost the speeds” followed by “alternate law protections”, recognizing the change in Airbus flight control laws.

Before we continue, let’s pause to understand what’s happening to Flight 447. The Airbus A330 has three separate pitot tubes that measure airspeed via three independent Air Data Computers.

These pitot tubes are small, cylindrical probes that stick out from the aircraft’s fuselage, near the nose. They have a forward-facing hole that allows oncoming air to enter. As the aircraft moves forward, air is “rammed” into this tube, creating pressure that sensors can measure. This pressure is compared with static air pressure to calculate the aircraft’s airspeed.

This redundancy of having three separate pitot tubes linked to three separate computers is designed to protect against any single component failure. However, if all three pitot tubes encounter the same environmental conditions at the same time, like what Air France 447 was experiencing at that very moment – they can all fail together.

Flying through the violent thunderstorm cell with its dense concentration of ice crystals known as graupel, the pitot tubes have become blocked by this graupel. Despite the pitot tubes’ heating systems, the ice is accumulating faster than it can melt and drain in these particular atmospheric conditions. Such blockages typically clear within one to two minutes as heating elements work, a temporary situation pilots are trained to manage using established procedures.

As these blockages disrupt the pressure readings, each system begins reporting different, fluctuating airspeed values – causing the aircraft’s computers to receive contradictory information.

When two or more of these systems provide conflicting data, the aircraft triggers an “unreliable airspeed indication” – exactly what First Officer Robert meant when he reported “we’ve lost the speeds.”

Unknown to the passengers sleeping peacefully in the cabin, Air France had recently begun upgrading to newer, more robust pitot probes across their fleet following reports of similar icing incidents. For now, though, this aircraft continues its journey through the night sky with the original less efficient pitot tubes still in place.

This inconsistent data triggers a critical change in the Airbus’s flight control system. The system switches from “Normal Law” to “Alternate Law,” fundamentally altering how the aircraft responds to pilot inputs. In Normal Law, Airbus’s fly-by-wire system acts as a safety measure, preventing pilots from exceeding safe operating parameters for example by being too aggressive with their control inputs. Now in Alternate Law, several automatic protections are disabled. Most crucially, stall protection is no longer available, meaning the aircraft won’t automatically prevent dangerous pitch up movements. The aircraft is now in a less automated, more manual mode meaning that the pilot flying needs to be very careful with his movements of the sidestick.

One last important point to mention here is that Airbus’s separate sidesticks don’t move in unison like Boeing’s linked control columns, meaning the pilots can’t physically feel each other’s control inputs – and this fact is about to become a monumental communication challenge for the flight crew of Flight 447.

Not only that, but if both pilots make inputs on their sidesticks simultaneously, something unexpected happens – the inputs get averaged out. So, for example, if one pilot pulls back on the sidestick, and the other pushes forward to the same extent, the net effect is that the inputs cancel each other out and nothing would happen.

Still only mere seconds after the autopilot disconnected, Robert is trying to work through the aircraft’s automated messages to figure out what’s happening. Adding to the confusion, the flight directors – providing visual cues for pitch and roll on the pilots’ displays – initially disappeared when the autopilot disconnected, but then reappeared shortly after. They began displaying guidance based on incorrect information, potentially suggesting pitch-up commands at the very moment when the proper response would have been to pitch down.

Robert says, “Wait we’re losing…” followed by “wing anti-ice”, suggesting concern about icing conditions.

While Bonin is still making pitch adjustments, Robert is concerned about their speed dropping as a result of the climb and calls out: “Watch your speed. Watch your speed.”

Bonin responds, “Okay, okay, okay. I’m going back down.”

“Stabilize,” Robert instructs, followed immediately by an acknowledgment from Bonin: “Yeah.”

Robert’s concern about their climbing trend continues: “Go back down.” Bonin’s instrument panel is showing the aircraft climbing above their assigned cruising altitude.

Robert: “According to that, we’re going up.” Adding with urgency, “According to all three, you’re going up, so go back down.”

“Okay,” Bonin acknowledges, but unfortunately his sidestick inputs don’t match his verbal confirmation.

“You’re at…Go back down,” Robert repeats, watching their altitude increase.

“It’s going, we’re going back down,” though the aircraft’s trajectory doesn’t change substantially.

“Gently,” Robert cautions, possibly noting Bonin’s aggressive control inputs.

Robert offers, “I’ll put you in in A.T.T,” most likely referring to attitude mode, believing it might help stabilize the aircraft.

“What’s that…”, Robert asks, showing his confusion about the aircraft’s behavior.

Bonin confirms, “We’re in… yeah, we’re in climb,” as they continue ascending past their assigned altitude.

At this point, a mere 46 seconds have elapsed since the first sign of trouble began when the autopilot disconnected.

First Officer Robert calls for the Captain by pressing a call button that sounds in the crew rest room: “Where is he?” he utters, exclaiming it with urgency, showing signs of stress. Robert begins calling for Captain Dubois repeatedly, asking “is he coming or not?”

The situation now escalates dramatically. After briefly seeming to stabilize with pitch around 6 degrees and angle of attack below 5 degrees, the stall warning triggers again. This time it sounds continuously, accompanied by the “cricket” sound specifically designed to grab attention.

Meanwhile, in the cabin, passengers and crew were most likely beginning to sense something wasn’t right. A female voice is heard saying “hello” a couple of times on the interphone, indicating cabin crew are becoming aware of the abnormal situation. Who knows what thoughts were racing through their minds as they felt the erratic movements of the aircraft.

Back in the cockpit, vibration noises can now be heard as the airframe begins to buffet due to disturbed airflow.

At this point, we’re still only approximately 60 seconds into the ordeal. Robert urges: “Above all try to touch the lateral controls as little as possible eh.” This is proper procedure for unreliable airspeed – minimize roll inputs to maintain stability.

The stall warning persists again: “Stall, stall,” with Bonin responding “I’m in TOGA eh”, indicating he believes maximum power should resolve their situation. This contradictory response – applying power while pitching up – reveals a fundamental confusion about the aircraft’s energy state. He should have lowered the nose instead of raising it.

During this time, the Trimmable Horizontal Stabilizer (THS) begins to move, gradually shifting from 3 degrees nose-up to an eventual 13 degrees nose-up position over the next minute. This trim movement significantly compounds the difficulty of recovery.

As the aircraft continues to climb, the angle of attack continues increasing as Bonin maintains his nose-up inputs. The roll oscillations begin to develop as the aircraft reaches its maximum altitude of 38,000 feet. The pitch attitude and angle of attack both reach 16 degrees, well beyond the critical angle for sustained flight at this altitude.

The angle of attack, which is the angle between the wing and the oncoming airflow, had been gradually increasing when the stall warning sounds again. As angle of attack increases, lift initially increases and this is essentially how aircraft climb. However, there’s a critical angle at which the airflow can no longer follow the contour of the wing and separates, causing a dramatic loss of lift. This is a stall, and it can happen at any speed if the critical angle of attack is exceeded.

In cruise flight at high altitude, the air is thinner, providing less lift. The margin between normal cruise speed and stall speed is narrower (known as the “coffin corner”), and recovery requires more altitude to regain airspeed. Most pilot training before this incident focused on stall recognition and recovery at a relatively low altitude, not at cruise altitude.

Over blaring alarms and alerts, Robert confusingly asks, “But we’ve got the engines, what’s happening? Do you understand what’s happening or not?”

Bonin doesn’t answer for 7 seconds until he finally admits: “I don’t have control of the airplane anymore now,” followed by, “I don’t have control of the airplane at all.”

Approximately 90 seconds have now elapsed since this unfortunate sequence of events began. Bonin’s response, pulling back on the sidestick when faced with unreliable airspeed might seem puzzling, but human factors help explain this reaction.

The “startle effect” is a physiological response to unexpected stimuli. When suddenly confronted with multiple alarms, unusual aircraft behavior, and confusing instrument readings, pilots can experience increased heart rate, muscle tension, narrowed perceptual field (tunnel vision), and difficulty processing complex information. Hearing is often one of the first senses to deteriorate. They also tend to revert to more familiar, ingrained responses and that seems to be what might have been happening to Bonin here. When Airbus aircraft are in Normal Law (as they usually are), the standard response to unexpected turbulence or discomfort at cruise is often to climb slightly for a smoother ride. Bonin’s instinctive pull-back response may have been this type of habitual reaction, inappropriate for the actual situation but perhaps triggered by stress and confusion.

“Cognitive tunneling” further compounded the problem. This psychological phenomenon causes people under stress to fixate on certain information while missing other critical cues. The pilots became focused on controlling the aircraft’s roll and managing the immediate crisis of lost airspeed, while missing the developing stall condition despite physical cues like airframe buffeting and the stall warning sound.

Once Bonin admitted that he no longer had control of the airplane, it took Robert an agonising 3 seconds to request “controls to the left” as he attempted to take control. The proper procedure for transferring control requires a formal “taking control” callout and pressing the priority button. Neither pilot follows this procedure completely.

To add to the already heightened level of confusion in the cockpit, without saying anything, Bonin almost immediately takes back control again and continues his nose-up inputs.

A few seconds later, Robert exclaims, “What is that?” still struggling to understand the aircraft’s behavior. “I have the impression we have… the speed,” Bonin responds.

With the aircraft now in a fully developed stall, its behavior became increasingly difficult to manage.

At this point, Captain Dubois returns to the cockpit and immediately asks, “Er, what are you doing?” as he observes the chaotic situation. It might seem like it took a considerable length of time for the captain to show up, and probably felt like forever to Robert who was the one calling him. But in reality, it was only a little over 90 seconds since the autopilot got disconnected and less than one minute from when Robert started pushing the call button.

Robert: “What’s happening? I don’t know I don’t know what’s happening”

Bonin: “We’re losing control of the aeroplane there”

Robert: “We lost all control of the aeroplane we don’t understand anything we’ve tried everything”

By this time, all recorded speeds have become invalid again, and the stall warning has stopped – not because the stall condition has been resolved, but because the system invalidates the angle of attack values when measured airspeed falls below 60 knots.

The aircraft is now descending rapidly at a staggering 10,000 feet per minute. The angle of attack exceeds 40 degrees, well beyond the critical stall angle. The engines are at nearly 100%, producing maximum thrust, yet the aircraft is descending due to the extreme angle of attack creating overwhelming drag.

The aircraft develops roll oscillations that reach up to 40 degrees. Completely disoriented, Bonin holds his sidestick at the left and full nose-up stop for about 30 seconds – inputs that only worsen the stalled condition.

What follows is a few more seconds of confusion as the pilots try to make sense of the situation.

Bonin: “I have a problem it’s that I don’t have vertical speed indication.”

Dubois: “alright”

Bonin: “I have no more displays”

Robert: “We have no more valid displays.”

At this moment, the thrust levers are at IDLE, though the engines’ N1 remains at 55%.

It’s plausible that Bonin now considers that their aircraft might be at risk of overspeed when he repeats his earlier observation, “I have the impression that we have some crazy speed no what do you think?”

Perhaps he reached out for the speedbrakes after saying that because Robert quickly warns, “No above all don’t extend” followed again by “don’t extend”. Bonin then makes pitch-down inputs, momentarily decreasing the angle of attack. The stall warning briefly reactivates with “stall, stall” as the airspeed momentarily increases above 60 knots due to Bonin’s slight forward sidestick movement.

Whenever back pressure was maintained on the sidestick, the angle of attack remained high, but the airspeed was so low that the stall warning was silent. This is because the Airbus stall warning system was designed to deactivate at extremely low airspeeds when angle of attack data is considered unreliable. When they momentarily pushed forward, the correct recovery action, the airspeed would begin to increase just enough to reactivate the stall warning system. This created the counterintuitive impression that pushing forward was causing the warning, while pulling back was silencing it.

In this unusual situation of a deep stall at high altitude, the system essentially warned them when they were doing the right thing and stopped warning them when they were doing the wrong thing, an incredibly confusing situation for the pilots in an already high-stress environment.

Bonin then announces, “So we’re still going down,” finally recognizing their massive descent rate.

Robert responds, “We’re pulling,” as they attempt to stop the descent.

A second later, Robert astonishingly asks, “What do you think about it? What do you think? What do we need to do?” showing how completely the situational awareness has broken down.

The Captain, clearly just as confused, responds, “There, I don’t know, there it’s going down.”

5 seconds later, Bonin, says, “That’s good we should be wings level, no it won’t,” showing concern about the aircraft’s bank angle.

The Captain advises, “The wings to flat horizon, the standby horizon,” trying to get Bonin to use their backup attitude indicator to level the wings.

Robert adds: “The horizon (second),” possibly acknowledging the standby horizon.

By now, almost 2 and a half minutes into their ordeal, they’ve passed through FL 250 (25,000 feet), having lost over 13,000 feet of altitude in less than a minute with their vertical speed indicating more than -15,000 feet per minute at times.

The confusion continues with conflicting perceptions of the aircraft’s attitude.

Robert: “You’re climbing” followed immediately by “You’re going down down down.”

Bonin: “Am I going down now?”

To which Robert and Dubois respond with conflicting answers.

Robert: “Go down”

Dubois: “No you climb there.” “You’re climbing!”

Bonin: “I’m climbing okay so we’re going down”

More stall warnings, the cricket chime and the continuous C-chord alert.

Bonin: “okay we’re in TOGA,” referring to the thrust setting. “What are we here?” followed by “On alti what do we have here?” and repeating that same question seconds later showing that he’s trying hard to determine their altitude.

In apparent disbelief, the captain utters “it’s impossible”.

Robert asks “what do you mean on altitude?”

Bonin: “Yeah yeah yeah I’m going down, no?”

Robert: “You’re going down yes.”

Dubois: “Hey you.” “You’re in…” but doesn’t complete the thought.

Dubois: “Get the wings horizontal”

Robert echoing Dubois: “Get the wings horizontal”

Bonin: “That’s what I’m trying to do”

Dubois repeats himself: “Get the wings horizontal”

Bonin: “I’m at the limit…with the roll”

The dual input warning then sounds indicating that both Bonin and Robert are making simultaneous inputs on their sidesticks.

The Captain mentions “The rudder bar,” suggesting use of rudder input to help control the aircraft, “Wings horizontal…go… gently gently,”

Even more frustration follows as Robert declares, “We lost it all at left”, “I’ve got nothing there” possibly referring to loss of control authority.

Bonin remarks, “We’re there we’re there we’re passing level one hundred” – meaning 10,000 feet.

The crew has now been struggling with the situation for just over three minutes, with multiple control transfers and conflicting inputs throughout the descent.

Finally, Robert says, “Wait me I have I have the controls eh,”

Another dual input announcement showing that two pilots are again attempting to control the aircraft at the same time.

Bonin: What is… how come we’re continuing to go right down now?”

Robert: “Try to find what you can do with your controls up there”. “The primaries and so on,” presumably referring to primary flight controls.

At this point, the Captain responds “do anything!” followed by, “it (won’t do) anything” , indicating his growing desperation realising that his co-pilots’ control inputs are no longer effective.

Robert calls out “Nine thousand feet” as they continue with their rapid descent.

Dubois: “Careful with the rudder bar there,” again warning about excessive rudder inputs.

Robert: “Climb climb climb climb,” apparently still not recognizing the stall condition they were faced with.

Bonin responds: “But I’ve been at maximum nose-up for a while!” revealing he’s been maintaining the worst possible input during a stall situation.

Another dual input warning sounds as Dubois says, “No, no, no, don’t climb.”

Robert: “So go down” followed by “So give me the controls… the controls to me controls to me”, trying to take definitive control of the aircraft.

Bonin at last says, “Go ahead you have the controls we are still in TOGA eh,” formally transferring control to Robert.

Dubois: “(so wait) AP OFF”

Dubois: “Watch out you’re pitching up there”

Robert: “I’m pitching up”

Dubois: “You’re pitching up”

Robert: “I’m pitching up”

Bonin: “Well we need to we are at four thousand feet”

Dubois: “You’re pitching up”

The ground proximity warning system urgently announces: “sink rate”, “sink rate” and then “pull up” warnings.

Dubois: “Go on pull”

Bonin: “Let’s go pull up pull up pull up”

Bonin: “we’re going to crash! This can’t be true.” “But what’s happening?”

Dubois: “(ten) degrees pitch attitude”

Those were the last words ever uttered by the flight crew.

At 02:14 am and 28 seconds, Air France Flight 447 strikes the ocean surface at a vertical speed of 10,912 feet per minute, a ground speed of 107 knots, and a nose up attitude of 16.2 degrees.

There were no survivors.

In the aftermath of the crash, investigators faced a seemingly impossible task. Air France Flight 447 had vanished into one of the deepest parts of the Atlantic Ocean. After three unsuccessful search campaigns, it wasn’t until April 2011, almost two years later, that the wreckage and flight recorders were finally discovered.

Coincidentally, the earlier ‘Titanic of the skies’ metaphor seems more apt as it was Woods Hole Oceanographic Institution, the same organization that found the actual Titanic, that led the successful discovery of the aircraft wreckage approximately 6.5 nautical miles northeast of the last known position.

The investigation, conducted by France’s BEA, revealed a sequence that began with a common occurrence: the temporary obstruction of the Pitot probes by ice crystals known as graupel. When the airspeed readings became unreliable, the autopilot and autothrust systems automatically disconnected as designed.

What should have followed was a straightforward procedure: maintain attitude and thrust, and wait for reliable speed indications to return. Instead, the flight data recorder revealed that First Officer Bonin maintained persistent nose-up inputs for over three minutes. His grip on the sidestick remained firm, keeping a nose-high attitude despite the aircraft’s increasingly precarious position. The investigation would later determine that these sustained inputs became a critical factor in the developing situation.

Several other factors contributed to the crash:

First, the startle effect of the unexpected autopilot disconnection at cruise altitude degraded the crew’s capacity for rational analysis.

Second, the crew had received insufficient training for handling unreliable airspeed situations at high altitude, where aerodynamic characteristics differ significantly from low altitude.

Third, no clear display indicated “unreliable airspeed,” making diagnosis more difficult amid multiple ECAM messages.

Most critically, the crew didn’t recognize the aircraft’s approach to stall, despite warning sounds and buffet. In the A330’s alternate law, the stall warning is harder to interpret at high altitude, where the margin between normal cruise angle of attack and stall angle is much smaller.

The Air France Flight 447 crash revealed critical gaps in pilot training and aircraft systems, prompting industry-wide reforms. The 228 souls aboard weren’t just passengers and crew. They were parents, children, partners, and friends. Their legacy lives on in every procedure rewritten and pilot retrained, ensuring countless families reunite safely at airports worldwide.

📄 SOURCES
Final report: https://finalreports.net/2009/air-france-447.pdf