Motl's blog notes:
> Anna has suggested that their GPS-based timing device may have
> neglected that the electromagnetic waves are moving slower than c
> through the atmosphere: if the collaboration did an error in this
> subtlety, they get an error of exactly the same magnitude to explain
> the "signal". The index of refraction of the air is 1.0003, so light
> needs to penetrate a 10-km layer of the atmosphere as it would need
> to get through 10.003 km of the vacuum which would exactly produce
> the 3-meter delay. Make the atmosphere a bit thicker because the
> satellites are not right above your head; add the delays from both
> directions and you may already produce those 18 meters of error (or
> most of it).>
> I blame frame-dragging myself... :)
> Damien