FROM THE MAGNETISM OF THEIR BALANCES. 361 



done, by placing the chronometer, when on board, in the same 

 position that it occupied on shore, when its rate was taken, and 

 turning it whenever the ship's course was changed. On board 

 vessels navigating in trade-winds, and in all voyages where a 

 steady course is preserved for a long time together, this would 

 not be difficult to accomplish ; but as a good deal of trouble 

 would attend this management, in some voyages, it appears to 

 me that it would not be impracticable to fix a pocket chrono- 

 meter on the top of a thin plate of metal or wood, suspended 

 on a needle-point as a centre, and moved by a magnetised bar. 

 To diminish, as far as practicable, the influence of the magne- 

 tism of this bar over the chronometer, the plane for the sup- 

 port of the chronometer might be fixed a few inches above the 

 needle ; and to prevent error from agitation, the apparatus 

 could be fixed on gimbles like a compass. 



It may be objected, that the influence of the magnetic bar, 

 connected with the apparatus for carrying the chronometer, 

 would induce magnetism in the balance, notwithstanding it 

 might be at some inches distance, and thus augment the source 

 of error. 1 apprehend, however, that as the bar is proposed to 

 be placed beneath the chronometer, so that its action would 

 be almost vertical to the plane in which the balance vibrates, 

 very little, if any effect, would be produced on the rate of the 

 instrument. Besides, were the rate of the chronometer taken 

 on shore when fixed in a certain position on this apparatus, 

 there would be every chance of its maintaining its rate at sea, 

 while the dip was nearly the same ; for the action of terrestrial 

 magnetism, combined with that of the local attraction of the 

 ship, would produce a mean action on the bar carrying the 

 chronometer, and a similar action on the chronometer. Un- 

 der great changes, indeed, in the magnetic intensity or dip, a 

 chronometer even thus situated might be liable to a small va- 



z z 2 riation 



