COMPARISON OF TIME. 35 



are repeated at every two seconds for at least one full minute. The opera- 

 tion is then reversed by the observer at the second station making the 

 breaks which are recorded at both stations as before. 



The differences of time between the chronometers at the two stations 

 are read from the chronograph sheets at each station and corrected for 

 error of the chronometers. The results from the two chronograph sheets 

 will differ by an amount equal to twice the time occupied in transmission 

 of signals. The mean of the two is therefore the approximate difference of 

 longitude. 



This result is yet to be corrected for personal equation, or the differ- 

 ence between the errors of observing of the two observers. Every observer 

 has the habit of recording a transit a little too early or too late, the differ- 

 ence between two observers not infrequently being as great as a fourth of 

 a second. To measure this difference, the observers usually meet, prefera- 

 bly at the known station, both before and after the campaign, and observe 

 for time each with his own instrument, or with one similar in all respects 

 to that used in the campaign. A comparison of the time determinations 

 made by the two observers gives an approximation to the personal equation. 



A better method, but one not always practicable, is for the observers, 

 having completed half of the observations for time and longitude, to ex- 

 change stations for the remainder of the work. The mean of the results 

 before and after exchange of stations will eliminate personal equation. 



There is one error incident to this work which can not be eliminated. 

 This is the unequal attraction of gravity, or local attraction, or, as it is 

 sometimes called, station error. The neighborhood of a mountain mass 

 will attract the plumb line and deflect the spirit level to such an extent as 

 to cause serious errors in astronomical determinations of latitude and time. 

 The same result is frequently produced by a difference in density of the 

 underlying strata of rock, so that station errors of magnitude often appear 

 where they are not expected. Indeed, the station error can not be pre- 

 dicted with any certainty, either as to amount or even direction. 



The only practical method of even partially eliminating this error is 

 to select a number of stations for astronomical location, under conditions as 

 widely diverse as possible, connect them by triangulation, and by this 



