[ 83 ] 
very inaccurate, if they are at any conhderable diflance 
one from the other; or if they were near, the two nee- 
dles would influence one another, and neither of them 
could be true : nay, fuppofing them for once to be pro- 
perly placed in this line, the leaf! motion of the fliip 
throws them out again. But this inftrument has a con- 
ftant power in itfelf, not only of fetting itfelf in the pro- 
per polition, but alfo of keeping itfelf fo ; or of reftoring 
itfelf to the fame fituation, if at any time it has loft it ; 
and it is curious to fee how, by its double motion, it 
counteradts, as it w'ere, the rolling motion of the veflel. 
I have only one thing farther to obferve, that as it is im- 
poflible for human hands to make any inftrument ma- 
thematically true, fo when we have two graduations to 
look to, as in the prefent cafe, one on the North, and the 
other on the South of the needle, we ought to attend to 
bothj and take the medium for the true dip or variation 
pretty nearly. But in this compafs there is another me- 
thod of examining the obfervations. Take a good arti- 
ficial magnet, and on the outfide of the compafs-box, 
point one end of it towards the needle, and by moving 
your magnet you may thus guide the north-end of the 
needle round to the fouth ; or, vice verfd^ without open- 
ing your compafs-box. The magnet being then laid afide, 
the needle will come to its true polition, after a few vi- 
brations ; but as both the needle and the circle of incli- 
nation are now reverfedrv, it will not poini: exadlly to the 
(h) Mr. iORiMER means, that the magnet fliould be applied in fuch man- 
ner as to turn the parallelogram and «ircle of inclination half way round hori- 
zontally, fo that that end of the axis of the needle which before pointed to the 
weft, fhall now point to the eaft. 
M 2 fame 
