According to simple
emission theory,
light thrown off by an object should move at a
speed of c with respect to the emitting object. If there are no complicating
dragging effects, the light would then be expected to move at this same speed until it eventually reached an observer. For an object moving directly towards (or away from) the observer at v, this light would then be expected to still be travelling at (c+v) ( or (c-v) ) at the time it reached us. In 1913,
Willem de Sitter argued that if this was true, a star orbiting in a double-star system would usually, with regard to us, alternate between moving towards us and away from us. Light emitted from different parts of the orbital path would travel towards us at different speeds. For a nearby star with a small orbital velocity (or whose orbital plane was almost perpendicular to our line of view) this might merely make the star's orbit seem erratic, but for a sufficient combination of orbital speed and distance (and inclination), the "fast" light given off during approach would be able to catch up with and even overtake "slow" light emitted earlier during a recessional part of the star's orbit, and the star would present an image that was scrambled and out of sequence. That is,
Kepler's laws of motion would apparently be violated for a distant observer. De Sitter made a study of double stars and found no cases where the stars' computed orbits appeared non-Keplerian. Since the total flight-time difference between "fast" and "slow" light signals would be expected to scale linearly with distance in simple emission theory, and the study would (statistically) have included stars with a reasonable spread of distances and orbital speeds and orientations, de Sitter concluded that the effect
should have been seen if the model was correct, and its absence meant that the emission theory was almost certainly wrong. ==Notes==