To make one determination of a star's place comparable with another made at another time, it is clear that we must correct for the effects of precession in the interval of time between the two observations, and for the effects of refraction. But observations made with the transit circle must also be corrected for errors in the instrument itself. The astronomer will see to it that his instrument is made and is set up as perfectly as possible. The pivots on which it turns must be exactly on the same level; they must point exactly east and west, and the axis of the telescope must be exactly at right angles to the line joining the pivots in all positions of the instrument. These conditions are very nearly fulfilled, but never absolutely. Day by day, therefore, the astronomer has to ascertain just how much his instrument is in error in each of these three matters. Were his instrument absolutely without error to-day, he could not assume that it would remain so, nor, if he had measured the amount of its errors yesterday, would it be safe to assume that those errors would not change to-day. In the examination of these sources of error the mercury trough comes again into use. The transit circle is turned directly downwards, and the mercury trough brought below it. A light is so arranged as to illuminate the field of the telescope, and the observer, looking in, sees the ten transit wires and the one declination wire, and also sees their images reflected from the surface of the mercury. If the telescope be pointing _exactly_ down towards the surface of the mercury, then the image of the declination wire will fall exactly on the declination wire itself, and by reading the circle we can tell where the zenith point of the circle is. Similarly, if the pivots of the telescope are precisely on the same level, the centre wire of the right ascension series would coincide with its reflected image. A third point is determined by looking through the eye-piece of the north collimator telescope--that is to say, the telescope mounted in a horizontal position at the north end of the room--at the spider lines in the focus of the south collimator. In order to get this view, the transit telescope has either to be lifted up out of its usual position, or else the middle of the tube has to be opened. The spider lines in the north collimator are then made to coincide with the image of the wires of the south collimator. The transit telescope is then turned first to one collimator, then to the other, and the central wire of the right ascension series is turned till it coincides with the wire of the collimator; the amount by which it has to be moved giving an index of the error of collimation; that is to say, of the deviation of the optical axis of the telescope from perpendicularity to the line joining the pivots.