Most easily seen by setting all the B elements to 1 and switching the B symmetry to 222. Some green vectors appear where they are certainly not perpendicular to the sphere radius. Since the color is determined by the dot product of the effective dipole vector and the radius, the part where the dot product is calculated must be having some problems. Heck, it occasionally produces a NaN result, which is bizarre.
It turns out this is related to the YYY color bug. Fixed by fixing an array size mismatch.