The moon or sun illusion is the apparent enlargement of celestial bodies near the horizon in comparison with their perceived size higher in the sky. This size change posed a problem for ancient astronomers, because it might imply that the sun changed its distance throughout the day. The preferred solution was to say that the image size was enlarged near the horizon by atmospheric refraction. The explanation is wrong: the effect of atmospheric refraction is to reduce the vertical diameter of the image and make the sun appear oval, and red. Despite better scientific education and the evidence of photographs, the refraction explanation remains popular today. Later astronomers such as Ptolemy (c. 170) and Cleomedes (c. 200) were aware that refraction was at best an incomplete explanation, and they suggested psychological or perceptual explanations. Cleomedes argued that size perception followed a geometrical model, and that perceived object size was determined by the angular size (retinal image size) and the perceived distance; the effect of the atmosphere was to make objects appear further away, and thus larger. Alhazen (c.1040) also used the geometrical model, but argued that it was the sight of the ground that made objects appear further away. The geometrical model remains prevalent today among psychologists, despite the fact that observers report that the moon/sun appears nearer on the horizon than higher in the sky. An alternative approach was promulgated by Berkeley (1709), who argued that size and distance were perceived independently, through various ‘cues’. On this argument the horizon moon is enlarged through cues such as relative size, texture gradients on the ground, aerial perspective, reddish colour, and looking straight ahead rather than up. This is a more promising approach, though agreement is needed on the size of these effects.