Schematic of the shift in fringe positions due to dispersion (the effect has been exaggerated for clarity). The vacuum (no dispersion) interferograms are plotted with solid lines; those dispersed by air with dotted lines. Top: dispersion shifts the point of zero optical path difference for a star, due to different amounts of air path in each arm of the interferometer (the effective optical path difference measured as if in vacuum). Middle: the dispersion shifts for stars of equal colors were equal and canceled; the measured separation is the same. Bottom: stars of unequal colors are shifted by slightly different amounts by dispersion, and the resulting measured separation is different. For very extreme color differences, the shift can be hundreds of microarcseconds. Not shown are the shape distortions to interferograms.