Graybeard - thank you, sir, for the pointer.
The antenna has greater than 6dBiC gain. So those who who declare steered and those who declare omnidirectional both have something on their side.
If I may riff on this a little, the dBiC specification basically refers to gain over a hypothetical purely isotropic circularly polarized antenna. Circular polarization avoids the Polaroid (tm) Sun glasses effect. Isotropic means the reference antenna pattern is a sphere. A pure hemisphere pattern would give exactly 3dBiC gain. So the antenna is somewhere betwen 60 degrees and 45 degrees beamwidth and aims itself roughly at the satellite. If they were using classical service they were sitting close to directly under the satellite per this link.
Classic services coverage - Support - Inmarsat If they were using SwiftBroadband service they were off at an angle of about 55 degrees from vertical relative to the satellite.
SwiftBroadband Coverage - Support - Inmarsat
So if they were using the classical Inmarsat modes they would not lose coverage in a flat spin until the equipment ceased to function. If they were on either the bird over Africa or the one South of Mexico then they'd have intermittent coverage that would depend on their spin rate.
Now can we determine which mode they used? If they support the Internet service I suspect it would be the SwiftBroadband service the antenna would be setup to support. Although being electronically steered it might be able to track the satellite adequately for communications through some pretty awesome gyrations if they were slow enough. If it was GPS aided tracking would be a slam dunk. If the integrated SatCom equipment that steers it is dumb enough then it'd lose track fairly easily with a high rate of spin.
I must admit that I was thinking "aimed" in terms of the antenna I was used to which had a much narrower beamwidth and tracked relatively reliably down to at least 7 degrees from the horizon. It used operator input for approximate position and aimed itself from there with a tracking algorithm that would uncomfortable once the gyros providing basic antenna stabilization became upset. But then, ships don't generally go into aircraft type spins. It was perfectly adequate for a ship in a pretty heavy storm.
If I had to put money on it I'd say that pretty close to 0214 and before 0215 the plane had met its fate.
Again, thanks for the pointer. It answered a lot of questions.
JD-EE