Technical analysis indicates with near certainty that the airplane turned south. We know this from Inmarsat’s second logged value—the burst-frequency offset. For the sake of simplicity, I will refer to this value as the “Doppler value,” because it includes, most crucially, a measure of radio-frequency Doppler shifts associated with high-speed movement in relation to satellite position, and is a natural part of satellite communications for airplanes in flight. Doppler shifts have to be predicted and compensated for by airborne systems in order for satellite communications to function. But the compensation is not quite perfect, because satellites—particularly as they age—do not transmit signals in precisely the way airplanes have been programmed to expect. Their orbits may tilt slightly. They are also affected by temperature. These imperfections leave telltale traces. Although Doppler-shift logs had never been used before to determine the location of an airplane, Inmarsat technicians in London were able to discern a significant distortion suggesting a turn to the south at 2:40 a.m. The turn point was a bit north and west of Sumatra, the northernmost island of Indonesia. It has been assumed, at some analytical risk, that the airplane then flew straight and level for a very long while in the general direction of Antarctica, which lay beyond its range.
After six hours, the Doppler data indicated a steep descent—as much as five times greater than a normal descent rate. Within a minute or two of crossing the seventh arc, the plane dived into the ocean, possibly shedding components before impact. Judging from the electronic evidence, this was not a controlled attempt at a water landing. The airplane must have fractured instantly into a million pieces. But no one knew where the impact had occurred, much less why. And no one had the slightest bit of physical evidence to confirm that the satellite interpretations were correct.