REPORT ON A LEVEL LINK. 



17 



thelatterdistancehavingbeen levelled over in the suinnierof 1837, 

 and the former in that of 1838. For the same reason it appears 

 much better to divide the distance into stages and finish them 

 one at a time, than to go over the whole in one direction, before 

 returning upon any part of it ; it being much more probable 

 that errors depending on the state of the atmosphere will balance 

 each other in the former than in the latter case. 



My own experience, therefore, leads to the conclusion, that 

 no levelling can be expected to give a correct result, unless it 

 be performed in opposite directions, and the mean of both re- 

 sults be taken ; instead of depending, as Captain Lloyd appears 

 to have done, on the consistency of separate sets of successive 

 readings. I have myself invariably found (as that gentleman 

 also did.) the agreement of these to be almost identical, both in 

 the going and in the returning series, notwithstanding the great 

 progressive difference of these two series of levels from each 

 other ; of which progression not the smallest trace is discover- 

 able in the separate columns of the same series, I have entered 

 the more minutely hito this subject, because I am not aware 

 that any one has described, or even noticed the existence of 

 such differences before ; and should feel much interest in reading 

 the statements of any experienced person who had been engaged 

 in a similar undertaking, and had conducted it with sufficient 

 care to render the law of the errors in any degree discernible. 



VOL. VII.— 1838. 



i 



