[ 562 ] 
me tne moil diflindt view of diHant objedfs,. which 
alfo I always ufed in all obfervations where the tele- 
fcope was not applied. The obferver fhould always 
be careful to examine the adjuftment of his quadrant 
in the day-time, when the horizon is to be feen 
clearly, and particularly in the afternoon, when he 
expedts to make an obfervation the enluing nights 
Should he have failed of making this examination the 
afternoon before an obfervation, he may examine 
what the error was, if any, the next morning; or 
he may make this examination by means of the 
Moon itfelf, in which cafe, it will be beft to ufe the 
telefcope; though the horizon of the fea is, in general, 
by far the beft objedt for this purpofe. As I found 
my quadrant would feldom continue exactly adjuHed 
for twenty-four hours together, inHead of frefh ad- 
juring it every day, I chofe rather to examine what 
the error of the adjuflnient itfelf was. This is done 
by turning the index of the quadrant, till the horizon 
of the fea, or the Moon, or any other proper, objedf, 
appears as one, and then the number of minutes, by 
which o on the index differs from o on the arch, is 
the error of the adjufhnent. If o on the index Hands 
advanced upon the quadrant, before or to the left- 
hand of o on the arch, that number of minutes is to 
be fubtradted from all obfervations; but if it Hands off 
the arch behind, or to the right hand of o on the 
arch, it muH be added to the obfervations. By exa- 
mining the error of the adjuHment in this manner, by 
at leaH three trials, and taking a medium of the re- 
fults, one can fcarce err above half a minute in deter- 
mining the exadt error of the quadrant ; whereas one 
may be miflaken a minute, or more, by a Angle trial. 
The 
