Ok, I just noticed something else. Misha, did you rotate the camera for a portrait type of shot in the second image?

Look what happens when the second image is rotated to the
same (camera) orientation as the first. The patterns of color
abberations are nearly identical. This makes me think that
it's not a natural effect (i.e. prism stuff).

I wonder what would happen if you were to take a few test
shots with the same camera/lens/film; one shot of a white
sheet of paper with a spot light source in a darkened room,
and another shot of a black background with the light source
shining on the camera itself, in order to induce
light leakage.

Once you've got your test shots scanned in, jack up the saturation
and look for that pattern of color in them.