196 THE PEAKS AND MOUNTAIN RANGES
the measurement of the altitude of a peak is the refraction of the atmosphere. A ray of light from a peak to an observer's eye does not travel along a straight line, but assumes a curved path concave to the earth. The ray enters the observer's eye— I quote from Colonel Burrard——-in a direction tangential to the curve at that point, and this is. the direction in which the observer sees the peak. It makes the peak appear too high. This refraction is greatest in the morning and evening, and least in the middle of the day; it is different in summer from what it is in winter. One of the great Himalayan peaks visible from the plains of India would appear, from observations with a theodolite made to it from the plains, to fall 500 feet between sunrise and the afternoon, and to rise again 800 feet before sunset; and even in the afternoon, when it would appear lowest, it would still be too high by perhaps 700 feet. This is obviously a very fruitful source of error, and the difficulty of determining the error is increased by the fact that the curvature of the ray varies with the rarefaction of the atmosphere. In the higher altitude, when the rarefaction of the atmosphere increases, the ray assumes a less curved path. All these possible sources of error due to the rarefaction of the
