the Royal Society’s Injlruments . 389 
vations made on Sept. 5. the variation with the needle, 
in its upright pofition, was 21.36 by the South end, and 
21.27 by the North; with the needle inverted it was 
2 1 .1 9 by the South end, and 2 1 .29 by the North. The 
mean of all four is 21.28, which is the true variation at 
that time and placer*;, and is 8' lefs than that fhevn in 
the Upright pofition of the needle by the South end, 
which is the end always ufed in obferving; fo that by 
this day’s experiment the error of the inftrument ap- 
peared to be 8'; but by a mean of the obfervations of 
this and two other days it came out 1 o'. Since that time 
the needle has been altered; and, at the time of the ob- 
fervations in 1775? the error was fo fmall as to be fcarcely 
fenfible. 
Great care was taken that the metal, of vdiich this 
variation compafs is compofed, fhould be perfectly free 
from magnetifm. There is a contrivance in it for lifting 
the needle from off the point, and letting it down gently, 
to prevent injury in carrying from one room to another. 
The inftrument is conftrudted nearly on the fame plan 
as fome made by the late Dr. knight, dhe principal 
difference is, that in his the pin which carried the needle 
was not fixed to the lower frame as in this, but to the 
box; the confetpuence of which was, that when the needle 
had fettled, and the box was moved to make the index 
on the needle point to the proper mark, it was again put 
(b) The quantity found by taking a mean of all the four numbers is evi- 
dently the fame as that got by taking a mean of the two firft and of the two 
laft, and taking a mean of thofe two means. 
F f f 2 
into 
