The Horizon Illusion

The horizon illusion has been known since antiquity, and is still not well explained, though it is unmistakeable

The full moon rising in the east at sunset appears huge and close, especially over a flat horizon, as does the red setting sun in the west. At midnight, the moon sailing high seems small and distant. Of course, the size of the moon as measured by the angle subtended at the eye is the same in both cases (except as slightly modified by refraction). The difference is in the mental perception of size. This is possibly the most commonly recognized optical illusion, although it is different from the usual ones. It has received extensive study. Minnaert discusses it at length in The Nature of Light and Colour in the Open Air, pp. 155-166. The effect is most striking seen over flat land, such as the Flanders coast or the Llano Estacado of Texas, and is weaker over the sea or in the mountains.

Related observations that may have a bearing on it include the perception that the vault of the sky is shaped like a flat dome (perhaps like the roof of the Mormon Tabernacle in Salt Lake), or that the ground appears saucer-shaped when seen from a balloon. The Top Hat illusion, in which a hat whose width of brim and height are equal appears to be much taller than wide, may also be related. Gauss found that the direction of view relative to the body had some effect (i.e., whether you are standing or lying). The zenith feels closer to us than does the horizon, so we might interpret an object subtending the same angle as larger on the horizon than in the zenith. No theory is really satisfying, and the explanation may involve several causes. The observation, however, is incontrovertible, and affords many moments of pleasure.

The illusion is not restricted to the sun and moon. Constellations viewed near the horizon appear larger, and stretched out horizontally. When the celestial sphere is projected stereoscopically on a plane, the distortion is similar. Planetariums seem to have hemispherical domes, which may be a mistake. It would appear more realistic if the dome were flattened.

The moon is estimated to appear from 2.5 to 3.5 times as large on the horizon as in the zenith. This is something you can investigate for yourself, comparing the moon with cardboard discs close to you, as described by Minnaert. The comparison must be made sequentially; if the moon is compared side-by-side with a disc, you will perceive the actual size, not the apparent size. This can be done at different altitudes, and a plot made of perceived size versus elevation. Such experiments bring out forcefully the the great difficulty of such investigations, where physical measurements do not reflect perception.

If the moon appears larger, then it seems to be magnified. Does this imply that you can see finer detail in the moon when it is on the horizon than when it is high? This is subject to test, but design of the experiment is not trivial! It is pleasant to be aware of the horizon illusion and the apparent shape of the sky, and to look for further effects. Minnaert says the sky seems to be 50 to 60 yards distant, as far as the estimation of size of objects is concerned, but that distant objects, such as radio antenna towers, can push the vault back beyond them.

My own feeling is that the horizon illusion involves the property of the visual sense that seeks for meaning in what is seen. It is often observed that we do not really "see" something until it is recognized. When the moon is in the sky, it can be compared only with the vast heavens, and appears more or less distant in accordance with our appreciation of apparent distance. When it is on the horizon, it is automatically compared with the distance to the horizon, or with objects seen on the horizon. Since the horizon is regarded as distant in most cases, and objects like houses on the horizon appear small, the moon is interpreted as all the larger. A similar effect occurs in many other connections, as with perspective. A normal-size person appears a giant if assumed to be located where telegraph poles are approaching the vanishing point, and not at the real distance.

For recent work on the illusion, and an interesting theory, see Prof. McCready's discussion in Theory of the Horizon Illusion.

Return to Optics Index

Composed by J. B. Calvert 1999
Last revised 2 August 2003