In this particular case, the disturbing cause is found in the action of our own atmosphere. The rays of light from the star are bent out of a perfectly straight course as they pass through the various layers of that atmosphere, layers which necessarily become denser the closer we get to the actual surface of the earth. Every celestial body therefore appears to be a little higher in the sky than it really is. This action is most noticeable at the horizon, where it amounts to about half a degree. As both sun and moon are about half a degree in diameter, it follows that when they have really just entirely sunk below the horizon they appear to be just entirely above it. It happens, in consequence, on rare occasions, that an eclipse of the moon will take place when both sun and moon are together seen above the horizon. It was a great matter to discover this effect of refraction. It was soon seen that it was not constant, that it varied with both temperature and pressure. It is, indeed, the most troublesome of all the hindrances to exact observation with which the astronomer has to contend; partly because of its large amount--half a degree, as has been already said, in the extreme case--and partly because it is difficult in many cases to determine its exact effect. The double observation with the transit circle gives us, then, the place in the sky where the star _appeared_ to be at the moment of observation, not its true place; to find that true place we have to calculate how much refraction had displaced the star at the particular height in the sky, and at the particular temperature and atmospheric pressure at which the observation was made. [Illustration: THE MURAL CIRCLE.] The transit circle is a comparatively recent instrument. In earlier times the two observations of right ascension and declination were entrusted to perfectly separate instruments. The transit instrument was mounted as the transit circle is, between two solid stone piers, and moved in precisely the same way. But the great six-foot wheel, which was made as stiff as it possibly could be, was mounted on the face of a great stone pier or wall, from which circumstance it was called the 'mural circle,' and a light telescope was attached to it which turned about its centre. This arrangement had a double disadvantage--that the two observations had to be made separately, and the mural circle, not being a symmetrical instrument, was liable to small errors which it was difficult to detect. Thus, being supported on one side only, a flexure or bending outwards of either telescope or circle, or both, might be feared.