Confronted with faulty instrument readings and alarms going off in the cockpit, the pilots of an Air France jetliner struggled to tame the aircraft as it went into an aerodynamic stall, rolled, and finally plunged 38,000 feet into the Atlantic Ocean in just 3 1/2 minutes.
But the passengers on that doomed Rio de Janeiro-to-Paris flight were probably asleep or nodding off and didn't realize what was going on as the aircraft fell nose-up toward the sea, the director of the French accident investigating bureau said after releasing preliminary black-box data on the June 1, 2009, disaster.
All 228 people aboard the Airbus A330 died.
The brief, highly technical report by the BEA contains only selective remarks from the cockpit recorder, offers no analysis and assigns no blame. It also does not answer the key question: What caused the crash?
But several experts familiar with the report said the co-pilot at the controls, at 32 the youngest of the three-man cockpit crew, Cedric Bonin, may have responded incorrectly to the emergency by pointing the nose upward, perhaps because he was confused by the incorrect readings.
The plane's external speed sensors, called Pitot tubes, have long been considered a likely culprit in the disaster, with experts suggesting they may have been iced over. And the BEA investigators found that two sets of instruments on the plane gave different speed readings, with the discrepancies lasting less than a minute.
Since the accident, Air France has replaced the speed monitors on all its Airbus A330 and A340 aircraft.
An official at Airbus said the aircraft's nose should have been pointed slightly downward to enable the plane to regain lift after it had gone into an aerodynamic stall.
"This is part of the general pilot training for any aircraft," said the official. He was not authorized to speak on that subject and asked not to be identified by name.
Other aviation experts concurred. In an aerodynamic stall, a plane most often loses lift because it is traveling too slowly, and begins to fall out of the sky. Pointing the nose downward enables the aircraft to pick up speed, gain lift and pull out of the stall.
Pulling the nose up is "an inappropriate way to respond" to an aerodynamic stall, said Paul Hayes, director of air safety for aviation consulting firm Ascend Worldwide Ltd. "He either misidentified what was happening or became confused."
He cautioned that Friday's report was brief and that it was still unclear how the series of events started.
The flight data recorder and cockpit recorder were dredged from the ocean in early May, along with some bodies.
They showed, in addition to inconsistent speed readings, two co-pilots working methodically to right the plane manually after autopilot stopped. Captain Marc Dubois returned from a routine rest to the cockpit amid what moments later became an irretrievably catastrophic situation.
After the plane went into a stall, warnings sounded, the autopilot and autothrust shut off as designed, and the co-pilot not at the controls "tried several times to call the captain back," the BEA report said. The captain returned one minute and 10 seconds later, when the plane had climbed to 38,000 feet.
"During the following seconds, all of the recorded speeds became invalid and the stall warning stopped," the report said, but added that the plane never came out of its aerodynamic stall.
"The airplane was subject to roll oscillations that sometimes reached 40 degrees," the report said. The engines never stopped operating and "always responded to crew commands," the BEA said.
"The pilots never panicked," BEA director Jean-Paul Troadec said on RTL radio, adding that they maintained professionalism throughout.
The passengers, he suggested, probably fell to their deaths without knowing they were doomed.
Dinner had been served and "you can imagine that most passengers were already asleep or nodding off," Troadec said. He said the cabin crew never contacted the cockpit to see what might be wrong.
"It seems they didn't feel more movements and turbulence than you generally feel in storms, so we think that till impact they did not realize the situation," said Jean-Baptiste Audousset, president of a victims' solidarity association, "which for the family is what they want to hear, they did not suffer."
He was among a group of representatives of families who met with BEA officials to be briefed on their findings.
At least one expert disagreed with the theory of a soft descent.
Data from the flight recorders shows the plane was falling almost 11,000 feet per minute (124 mph, or 200 kilometers per hour), its nose slightly tilted upward.
"Eleven-thousand feet a minute is a huge rate of descent," said Ronan Hubert, who runs the Aircraft Crashes Record Office in Geneva. "I would say some of the people on board would have lost consciousness."
The crew had feared turbulence, and more than eight minutes before the crash the co-pilot at the controls advised the cabin crew "you should watch out" for turbulence ahead. He said the plane could not climb out of the cloud layer where the turbulence was happening because it was not cold enough.
Turbulence caused the pilots to make a slight change of course, but was not excessive as the plane tried to pass through the clouds.
Four minutes later, the plane's autopilot and autothrust shut off, the stall alarm sounded twice and the co-pilot at the controls took over manual control. A second co-pilot, David Robert, 37, was also in the cockpit.
Pilots on long-haul flights often take turns resting to remain alert. After Dubois returned to the cockpit, he did not take back the controls.
Just over two minutes before the crash, Bonin is heard to say, "I don't have any more indications." Robert says: "We have no valid indications."
Michael Barr, who teaches aviation safety at the University of Southern California, said the atmosphere in the darkened cockpit would have been chaotic: lights flashing, loud alarms, frequent messages.
He compared the pilots to emergency-room doctors struggling with a sudden influx of seriously injured patients: They were bombarded with problems that they had to quickly prioritize.
On top of that, they were completely dependent on the information the plane's computers gave them.
"You have to rely on your instruments," Barr said. "That's why when the instruments aren't telling you the truth, you have a hard time deciding what to do. Which ones are right and which ones are wrong?"
Air France said in a statement that, based on the report, it appears "the initial problem was the failure of the speed probes which led to the disconnection of the autopilot" and loss of pilot protection systems.
The airline defended the captain, saying he "quickly interrupted his rest period to regain the cockpit."
Independent aviation analyst Chris Yates said the report appears "to raise more questions than it answers."
"It would seem to me, reading between the lines, that the cockpit crew weren't confident of the information that was being presented to them on the data displays," Yates said. "Maybe _ and it's only a maybe _ they took some action that led to the stall warning, and the plane stalling and then being unable to correct it."
A new, but not final, report with some analysis is to be issued in July.
Cecile Brisson and Frank Jordans and APTN in Paris, Joan Lowy in Washington and David Koenig in Dallas contributed to this report.