66 Mr. Cavallo’s Magnetkal 
endeavour proved ineffectual, for the needle conftantly fettled 
in the fame direction, without any fenfible variation. 
With a needle thus fufpended a variation compafs might be 
very eafily conftru&ed, and it would perhaps be more accurate 
than thofe commonly in ufe. For this purpofe the needle 
ought to be about three inches long* and the piece of looking- 
glafs ought to be fixed upon the index of an Hadley’s fextant, 
which muft be placed horizontally under the needle, with its 
edge or fiducial line in the meridian of the place, in order to 
obferve the daily variation of the needle. I have made only a 
rough model of fuch a variation compafs, and it feemed to 
anfwer very well. This coitftruClion appears to have the fol- 
lowing advantages over the common fort : i ft, the needle being 
cylindrical, and without a hole in the middle, would be left 
fubjeCt to have more than two poles, adly. The needle being 
{lender, its poles would fland more exaCtly in its axis, which with 
the common flat needles is feldom the cafe. 3dly, It will ap- 
pear, by a little confideration, that in this conftruClion there is 
no need of the needle’s center of motion keeping always in the 
fame invariable point, which renders the conftruftion both very 
eafy and very accurate: and, laftly, as the fextant may be 
placed at a confiderable diftance below the needle, and the reft 
of the frame may be made of any fize, there would be no ne- 
eeffity of placing any brafs or other metal fo near the needle as 
might affeCl it in cafe this metal had any rnagnetifm, which 
generally happens with brafs. 
In order to examine the rnagnetifm of divers fubftances, be- 
fides the above defcribed needle, I ufed to put a final! magnetic 
needle upon water, and then bring the fubftance to be exa- 
mined near it, or place the fubftance itfelf upon water, 
fometimes 
