But in the first position it was 2".3 east.
Hence the mean of two positions is 2".3 west. This small quantity may be corrected by the adjusting screw, particularly if furnished with a micrometer head, and a similar trial repeated.
It is in this manner that the transit instrument is level led at the Royal Observatory, and the result is registered once a week in the transit book, and published with the annual observations.
The next adjustment, if not previously attended to by the maker, is to cause the image of a star to be formed accurately on the micrometer wire, which should be placed exactly in the focus of the object glass. This requires a very careful examination. Both a star and the micrometer should be seen at the maximum of distinct ness at the same time. A double star, such as Castor, y Arietis, or 7 Virginis, should be chosen for this pu• pose. If, when these stars are seen distinctly double, the eye-glass requires to be drawn either in or out to improve the distinctness of the micrometer wire, it is a proof that the object glass itself requires to be drawn out a si milar quantity, that the image may be thrown at its pro per distance. It should be remembered, that this ad justment is precisely the same for either long or short sighted persons ; the alteration required for either is to be effected by the eye-glass only. The position of the micrometer wire should next be examined, to ascertain if it be parallel to the horizon. For this purpose, let any terrestrial object be bisected, and the instrument turned slowly in azimuth ; if the same object remain bi sected when at each extremity of the field of view, it is a proof that the wire is horizontal. If the circle does not move in azimuth, the vertical wire may be rendered perpendicular by a similar process. In each case, the wires are supposed to be perpendicular. Should this not happen to be the case, the horizontal wire may be exa mined by a star in the equator, which should remain bi sected during the whole of its passage, and any error in the vertical wire may be rendered of no importance, by Almost all the instruments above described, resemble each other in this common principle, that they determine the zenith distances of the heavenly bodies by a mean of two observations, one with the face of the instrument to the east, the other towards the west.
Ramsdcn's and Cary's instruments generally require some previous adjuttments of the microscopes, that is, each microscope is supposed to mark zero, when two points or dots, marked on the limb of the instrument, arc brought into a vertical position, by means of a plumb line passing over and bisecting each of them. As these dots are as permanent as the divisions themselves, the observer need not be anxious to change frequently the position of his instrument, because the error of collima tion once found will remain permanently the same, as long as the micrometer wires remain undisturbed. A long series of stars, therefore, may be observed with the face of the instrument in one position ; and at a very con siderable interval, if it should so happen, a correspond ing series may be obtained in the contrary position, and this permanency of the error of collimation, and the in creasing degree of accuracy with which it becomes known, is an advantage peculiar to these instruments.
In Troughton's circles, no previous adjustment of the microscopes is necessary ; but as the result supposes that every part of the instrument remains the same, they cannot be too often reversed ; and we would advise the observer to change their position every 2• hours, which is no objection to an instrument moving freely in azimuth.
But in whichever of these modes the instrument be. constructed, it will always possess the beautiful proper ty of pointing out to the observer, with considerable ac curacy, the maximum of its own probable error, which will be shewn by the discordance of the error of colli mation, as deduced from different stars, and it may be taken as a general rule, that half the extreme difference of a great number of observations, will be nearly the maximum of error to which any one determination is lia ble.