the mean Denfity of the Earth. 7 8 7 
greater variety in this interefting problem, and becaufe 
alio the furvey would be more eafily made, on account 
of the ground being more in view at each ftation than in 
the cafe of a hill, which generally hides more than half 
the compafs from the obferver. In computing the rela- 
tive altitudes of all the principal ftations, let the opera- 
tions be performed mutually both backwards and for- 
wards, that is, from both of every two objects, having 
for that purpofe obferved at each of them the vertical 
angle of the other, namely, both the angle of elevation 
and the angle of depreffion, and take the mean between 
the two computed differences of altitude; for this ex- 
cludes the neceffity of making the proper allowances for 
refraction, and for the curvature of the earth ; fince the 
effect of each of thefe is balanced and corrected by that 
of the counter obfervation. But as to thofe points in 
the fections which are far diftant from the obferver, 
and where great accuracy is required, it may be proper 
to make the allowance for refraction and curvature, as 
there is generally no back obfervation by which their 
effects may be balanced. Thefe are the chief hints 
which at prefent occur to me, befides the general infor- 
mation to be derived by the computer from the perufal 
of the modes of computation that have been defcribed in 
this paper. As to the furveyor, he will ftrike out other 
5 E 2 ' convenient 
