88 Magnetic Dip in the United States. 



lines is from N. 65° 1' W. to S. 65° 1' E. Computing the dip 

 from these data we obtain the differences in the last column 

 above. These differences are much less than those before found, 

 and it seems highly probable that in this vicinity the isoclinal 

 lines make a greater angle with the parallels of latitude than they 

 do in Ohio. Yet the above observations are all embraced within 

 less than two degrees of longitude, and are therefore insufficient 

 to determine with much precision the dependence of the dip upon 

 the longitude. I think it improbable that the inclination should 

 be as great as 24° 59' according to these observations ; yet ad- 

 mitting such to be the case, we still obtain considerable differen- 

 ces between the observed and computed dip. Are these differ- 

 ences to be regarded as errors of observation, or as errors of the 

 hypothesis of parallel, straight and equidistant isoclinal lines? 

 In order to answer this question, it is necessary to consider all 

 the possible sources of error in magnetic observations. 



The errors arising from the inclination of the magnetic axis of 

 the needle to the axis of figure, and from inequality in the weight 

 of the arms, as well as the zero error of the graduation, appear 

 to have been provided against in Prof. Locke's mode of observa- 

 tion. That arising from the excentricity of the axis of the nee- 

 dle in relation to the vertical circle on which the readings are 

 made, is not alluded to. This error in my instrument commonly 

 amounts to one or two minutes, and sometimes even to five or 

 more. It is corrected by readings at both extremities of the nee- 

 dle. Prof Locke makes no mention of having employed this 

 precaution, and his language on page 321, where he says " the 

 dip is determined by eight distinct readings of each needle," 

 would seem to imply that he did not attend to it. With a good 

 instrument, no great error would ordinarily arise from this source, 

 yet it might easily amount to 2' or 3'. 



A more considerable source of error is that arising from the 

 uncertainty of the readings themselves. A dipping needle will 

 seldom come to rest twice in the same position. This arises, not 

 from a change in the direction of the magnetic force, but from 

 friction on the axis of the needle. Prof Locke's observations 

 exhibit this fact in a striking light. The difterence of the read- 

 ings with the face of the instrument east and west, and in the 

 same position of the needle, is equal to twice the zero error. Now 

 as this error may be assumed to be constant, we obtain by a com- 



