VOL. LXVI.J PHILOSOPHICAL TRANSACTIONS. 57 



served both ways as before. The reason of this is, that the mean of the 

 observed dips, in these 4 situations, differs very little from the truth, though the 

 needle is not well balanced, and even though a great many other errors are com- 

 mitted in the construction of the instrument ; provided the needle is made 

 equally magnetical after the poles are reversed as before ; * and that the dif- 

 ference of the observed dip, in these 4 situations, is not very great. 



The error which is most difficult to be avoided, is that which proceeds from 

 the ends of the axis being not truly cylindrical. I before said, that the parts of 

 them which rest on the agate planes are always exactly the same. The instru- 

 ment is so contrived however thnt we may on occasion, by giving the axis a little 

 liberty in the notches by which it is lifted up and down, make those planes bear 

 against a part of the axis distant about -rlnr or -jV^h of an inch from their usual 

 point of bearing. Now I find that when the axis is confined, so as to have none 

 of this liberty, and when care is taken, by previously making the needle stand at 

 nearly the right dip, that it shall vibrate in very small arches when let down on the 

 planes ; that then, if the needle is lifted up and down any number of times, it 

 will commonly settle exactly at the same point each time, at least the difference 

 is so small as to be scarcely sensible ; but if it is not so confined, there will 

 often be a difference of 20' in the dip, according as different parts of the axis 

 rest on the planes, and that though care is taken to free the axis and planes 

 from dust as perfectly as possible, which can be owing only to some irregularity 

 in the axis. Also, if the needle vibrates in arches of 5 or more degrees, when 

 let down on the planes, there will frequently be as great an error in the dip. It 

 is true that the part of the agate planes which the axis rests on, when the vibra- 

 tions are stopped, will be a little different according to the point which the 

 needle stood at before it was let down ; which will make a small difference in the 

 dip as shown by the divided circles, when only one end of the needle is observed, 

 thougii the real dip or inclination of the needle to the horizon is not altered : 

 but this difference is by much too small to be perceived ; so that the abovemen- 

 tioned error cannot be owing to this cause. Neither does it seem owing to any 

 irregularity in the surface of the agate planes, for they were ground and polished 

 with great accuracy ; but it most likely proceeds from the axis slipping in the 

 large vibrations, so as to make the agate planes bear against a different part of 

 it from what they would otherwise do. I have great reason to think that this 

 irregularity is not owing either to want of care or skill in the execution, but to 

 the unavoidable imperfection of this kind of work. I imagine too that this in- 

 strument is at least as exact, if not more so, than any which has been yet 

 made. By a mean of all the experiments with several needles, the true dip at 



* It is easy to see whether the needle is made equally magnetical after the poles are reversed as 

 before, by counting the number of vibrations which it makes in a minute. — Orig. 

 VOL. XIV. I 



