Instrumental Error of his Horizon-Sector. 829 



Unfortunately the wire could not be displaced more than 

 15', at which angle (of depression) the reflected disc appeared 

 so foggy, indistinct, and even elliptical, that — instead of ob- 

 serving the lower or upper limb (or both), — the vertical dia- 

 meter, which appeared under an angle of not less than two 

 minutes, was bisected, vaguely no doubt, but as exactly as 

 could be estimated. No reference level was made use of, al- 

 though clearly proved in a succeeding attempt to be absolutely 

 indispensable. So many sources of inaccuracy led, as might 

 be anticipated, to most unsatisfactory results ; the error coming 

 out, not the same by both levels, as should have been the case, 

 but 30" by the one, and 38" by the other. An extra horizon- 

 tal wire, fixed just within the limits of the stop, would insure 

 complete success to this novel method*. 



The measurement by the graduated arcs of the sector of the 

 elevation of the disc, and of the depression of its reflection at 

 an angle sufficiently great to render the image well defined, 

 although uncertain in an instrument not subdivided beyond 

 5", was of necessity adopted. The inclination of the colli- 

 mator being increased to about 1°, by means of a wedge fixed 

 to the north end of the under surface of the stand, the latter 

 was glued to the plank. Fortin's level, attached as well to 

 the collimator as to the adjoining side of the Ys which sup- 

 ported it, served to indicate and rectify, during the subsequent 

 operations, the minutest deviation in the inclination of the 

 plank. After the lapse of a day or two, the sector being sta- 

 tioned south of the cistern, with its vertical wire bisecting the 

 disc, the elevation of the upper and lower limbsf of the disc, 



disturbing the direction of its axis, it follows that the line of sight was ele- 

 vated above the axis at the same angle at which it is now depressed below 

 it; its vertical range being BPR. Hence the axis must lie in the direction 

 of the level line PL, which bisects the angle BPR. 



Admitting the cylinder to be flexible, the line of sight would be elevated 

 above its axis ; but on depressing the cylinder by an angle equal to the 

 deflection, the deviated wire would again bisect alternately the direct and 

 reflected disc, and the line of collimation, adjusted as usual, would point 

 level. 



* The readiest plan of observation by this method may be briefly stated. 

 —The sector being about level, alter the inclination of the collimator until 

 either limb of the disc appears covered by the additional wire of the sector. 

 Then invert the cylinder of the sector; and if the wire now points above the 

 same limb of the reflected disc, diminish the inclination of the collimator 

 by half the difference, or increase it equally if the image appears above the 

 wire. 



When the upper one of the three horizontal wires of a telescope bisects 

 a low star, and the lower one (is moved to bisect at the same instant) its 

 reflected image, the middle wire, when placed equidistant from the other 

 two, will obviously point level. 



fin fact, the horizontal wire was pointed a second or two below the 

 upper limb, and the same quantity above the lower one. 

 Third Series. Vol. 2. No. 1 1. May 1833. 2 U 



