lioi/ul Society. 33 



ready and perlecdy exact method of determining the position 

 of the horizontal or zenith point on the hmb of a circle or ze- 

 nith sector. Its principle is the invariability with respect to 

 the horizon of the position assumed by any body of invari- 

 able figure and weight floating on a fluid. It consists of a 

 rectangular box containing mercury, on which is floated a 

 mass of cast-iron about twelve inches long, four broad, and 

 half an inch thick, having two short uprights or Y's of equal 

 height, cast in one piece with the rest. On these is firmly 

 attached a small telescope furnished with cross wires, or, 

 what is better, crossed portions of the fine balance spring of 

 a watch, set flat-ways, and adjusted very exactly in the si- 

 dereal focus of its object-glass. The float is browned with 

 nitric acid to prevent the adhesion of the mercury, and is pre- 

 vented from moving laterally by two smoothly polished iron 

 pins, projecting from its sides in the middle of its length, which 

 play freely in vertical grooves of polished iron in the sides of 

 the box. When this instrument is used, it is placed at a short 

 distance from tlie circle whose horizontal point is to be ascer- 

 tained, on either side (suppose the north) of its centre, and 

 the telescopes of the circle and of the collimator are so 

 adjusted as to look mutually at each other's cross wires (in 

 the manner lately practised by Messrs. Gauss and Bessel), first 

 of all coarsely, by trial, applying the eye to the eye-glasses 

 of the two instruments alternately, and finally by illuminating 

 the cross wires of the collimator with a lanthorn and oiled 

 paper, (taking care to exclude false light by a black screen 

 having an apperture equal to that of the collimator,) and making 

 the coincidence in the manner of an astronomical observation, 

 by the fine motion of the circle. The microscopes on the limb 

 are then read ofl", and thus the apparent zenith distance of the 

 collimating point (intersection of the wires) is found. The 

 collimator is then transferred to the other (south) side of the 

 circle, and a corresponding observation made "without reversing 

 the circle, but merely by the motion of the telescope on the 

 limb. The difference of the two zenith distances so read off" 

 is double the error of the zenith or horizontal point of the 

 graduation, and their semi sum is the true zenith distance of 

 the collimating point, or the co-inclination of the axis of the 

 collimating telescope to tlie horizon. 



By the experiments detailed in Capt. Kater's paper, it ap- 

 pears that the error to be feared in the determination of the 

 horizontal point by this instrument, can rarely amount to half 

 a second if a mean of four or five observations be taken. In 

 a hundrtd and fifty one single trials, two only gave an error 

 of two .seconds, and one of these was made with a wooden float. 



Ill 



