In practice the sextant is an instrument only six or eight inches in diameter. It is held in the right hand and the movable radial arm is adjusted with the left hand with the aid of a micrometer screw, and the reading of the scale is made accurate by the vernier arrangement. The ordinary observation--which every traveler has seen a navigator make from the ship's bridge just at midday--is carried out by holding the sextant in a vertical position directly in line of the sun, and sighting the visible horizon line, meantime adjusting the recording apparatus so as to keep the sun's limb seemingly in touch with the horizon. As the sun is constantly shifting its position the vernier must be constantly adjusted until the observation shows that the sun is at the very highest point. The instrument being clamped and the scale read, the latitude may be known when proper correction has been made for the so-called dip, for refraction, and where great accuracy is required for parallax. [Illustration: "TAKING THE SUN" WITH THE SEXTANT. The instrument is held in the right hand, and levelled at the horizon; the left hand manipulating the micrometer screw which adjusts the radial arm carrying the index mirror (at top of figure). The result is read on the Vernier scale (arc at bottom of figure) with the aid of the magnifying glass.] Dip, it may be explained, is due to the fact that the observation is made not from the surface of the water but from an elevation, which is greater or less according to the height of the bridge, and which therefore varies with each individual ship. The error of refraction is due to the refraction of the sun's light in passing through the earth's atmosphere, and will vary with the temperature and the degree of atmospheric humidity, both of which conditions must be taken into account. The amount of refractive error is very great if an object lies near the horizon. Everyone is familiar with the oval appearance of the rising or setting sun, which is due to refraction. With the sun at the meridian, the refractive error is comparatively slight; and when a star is observed at the zenith the refractive error disappears altogether.