[ 100 ] 



©Bfervations be as much depended upon as the fore 

 obfervations. The caufes of this feem to have been- 

 principally thefe two, the difficulty of adjufting the 

 back horizon glafs, and the want of a method of 

 directing the fight parallel to the plane of the qua- 

 drant. The back horizon glafs, like the fore-onej- 

 requires two adjustments : the firft, or common one, 

 difpofes it at right angles to the index glafs, when 

 the index ftands at (o) upon the arch ; which is ufu- 

 ally performed by fetting (o) of the index of the arch 

 of the quadrant by double the dip of the horizon of 

 the fea, and then holding the quadrant vertical with 

 the arch downwards, and turning the back-horizon 

 glafs about, by means of its lever or perpetual fcrew,, 

 till the reflected back horizon appears to coincide 

 with the fore-horizon feen directly. But this ope- 

 ration is fo difficult in practice with the back-horizon 

 glafs wholly filvered, except a fmali tranfparent flit 

 in the middle, as it has been ufually made, that few (if 

 any) perfons have ever received proper fatisfaction 

 from it. If the back-horizon-glafs was filvered in 

 every refpect like the fore-horizon- glafs (which it 

 ought to be) the upper part being left unfilvered, and 

 a telefcope was applied to it, perhaps this adjustment 

 might be rendered fomewhat eafier and more exact 5 

 but it could not even thus be made fo exact as the 

 adjuftment of the fore-horizon-glafs may, by making 

 ufe of the Sun's limbs. 



The fecond adjuftment of the back-horizon- 

 glafs, in the common conftruction of the quadrant, 

 is ftill more troublefome, fince it cannot be exe- 

 cuted without fetting the index 90 degrees off 

 the arch, in order to place the index- glafs par- 

 1 allet 



