Hours after Flight 370 vanished on March 8, Inmarsat pulled together a team of engineers thousands of miles away at its London headquarters for a marathon data-crunching session to help find the missing jet.

Their mission: piecing together the few signals picked up by an Inmarsat satellite to direct the search effort, which in the initial phase focused on the Boeing 777's flight path from Kuala Lumpur to Beijing.

After a week of study, the data revealed two plausible, if surprising, trajectories — one heading north into Central Asia and one south toward Antarctica.

The breakthrough, which helped dramatically narrow the search area, was based on research using the so-called Doppler effect. It is named after 19th-century Austrian physicist Christian Doppler, who explored how movement can alter a signal profile.

Gym and pizza breaks

After establishing a broad flight path, the team hunkered down to pinpoint a more accurate possible crash site.

"They worked together for six or seven days straight," said Inmarsat spokesman Chris McLaughlin, with the team unwinding at the in-house gym during breaks or fetching pizza to sustain the round-the-clock mission.

"What we discovered was that the northern projected path had no correlating pings appear on it, while for the southern projected path, we had frankly an absolute correlation," McLaughlin said.

The teams, meeting in the glass-coated Inmarsat headquarters that borders London's financial district, compared equivalent data from flights of other Boeing 777 jets in and out of the area to see where the Doppler effect would result in a pattern that matched the Flight 370 data.

The team's breakthrough came over the weekend, when the group was able to rule out the northern corridor and home in on the lower end of the southern axis, he said.

Conclusion altered search

The mathematical model played a crucial role in a search mission now in its third week, with no conclusive trace of any debris to match the aircraft. While countries from Australia to China have rushed planes and ships to the site thousands of miles west of Perth, Australia, the vast field and bad weather have hindered the search.

Authorities said that Inmarsat would keep working on its research to further whittle down the search area, with the engineer's findings so far the most tangible breakthrough in a mystery that has otherwise remained puzzling for a global audience transfixed by the disappearance of a large aircraft.

"That's the best they can do with the information available to them," said Robert Mann, a former American Airlines executive and aviation consultant based in Port Washington, New York. "That's the limitation, as you're forced to innovate."