Variation of the Magnetic Needle. 65 



great there as at Cambridge or Salem, it follows that the dif- 

 ferences observed in New York are not too great to be ac- 

 counted for by the diurnal motion alone, without resortmg 

 to the hypothesis of an irregular increase in the mean quan- 

 tity of the variation. It may also be observed that the 

 variation found at the same time and place with different in- 

 struments will frequently vary half a degree or a degree; 

 and, by changing the place of the instrument a few feet, 

 the same effect will sometimes be produced. This is more 

 particularly the case in compact places, when the observa- 

 tions are made from the windows or on the top of a building; 

 the nails and other iron used in constructing it, having fre- 

 quently a great effect on the position of the needle. Not- 

 withstanding the difficulty of obtaining the correct values of 

 the variation, it is of importance to ascertain at regular inter- 

 vals, as correctly as possible, particularly in this country, 

 where most of the boundary lines of lands are determined 

 by the compass. To assist in this object, I shall here give 

 an abstract of my observations made at Salem in the years 

 1805, 1808, and 1811. 



The observations in the year 1805 were made at a house 

 in Summer Street, Salem, with a theodolite, furnished with 

 a quadrant of altitude, telescope, &c. grarluated to minutes. 

 After making the usual adjustments, and fixing as nearly as 

 possible the quadrant of altitude, and the north point of the 

 needle at the commencement of the graduation of their re- 

 spective circles, I estimated the errors of these last adjust- 

 ments, and applied them respectively to the observed alti- 

 tudes and azimuths of the sun, in a similar way to the meth- 

 od of correcting for the index error of an observation made 

 with a quadrant of reflection. To ascertain these index er- 

 rors to a greater degree of accuracy, I generally took the 

 mean of ten observations of the needle, and three observa- 

 tions of the quadrant, before and after each set of observa- 

 tions. The instrument was placed within the house, at an 

 eastern window in the morning, and at a western in the 

 evening, at the distance of two or three feet from the wall, 

 (or farther when it could be done) in order to avoid as much 

 as possible the effect of the iron in the walls of the building. 

 Having obtained in this way the sun's true altiiude and mag- 

 netic azimuth, the true azimuth was calculated and the va- 



VoL. XVL— No. 1. 9 



