CHAP. XIV.] TELEGRAPHIC MERIDIAN DISTANCE 327 



the difference of the two times wiU be two seconds too much. 

 The mean of the two will therefore be correct, and half the 

 difference of the two will give the " retardation of the ware " 

 — a matter, however, purely of curiosity as far as our results 

 are concerned. The mean also eliminates " personal equa- 

 tion " in sending and receiving signals. In land lines the 

 retardation is about one-tenth of a second per thousand miles ; 

 with submarine cables it is larger, and varies. 



To eliminate all personal errors, the only satisfactory 

 method is for the observers to change ends. When this is 

 not possible, personal errors, when obtainable, should be 

 applied. These can only be determined by a considerable 

 series of observations for time, and the error of each observer 

 or the person chosen as standard being recorded and meaned. 

 The steadiness, and therefore the value, of the personal error 

 can thus be judged. 



Personal equation in sending and receiving signals can best 

 be determined at an observatory by means of a chronograph ; 

 it is usually very much smaller than that found in the sextant 

 observations of different observers. 



In tliis case a number of chronometers is not necessary. Only One 

 All that is wanted is one good time-keeper. If, however, two ^^*°^ 

 watches are at hand, it is not amiss to ascertain the Errors sary. 

 of each separately, and use them both in transmitting the 

 signals. 



A box-clu'onometer is the best for sending and receiving 

 signals by, and, if practicable, it may be a good plan to land 

 one, and let it stand in the telegraph office for a few days 

 beforehand to settle down, comparing the watch actually used 

 at sights with it before and after observations. 



Sights must be obtained on the day of sending the signals. Time to 

 and the latter should be transmitted at or about noon or exchange 



Signals. 



midnight. Where the places are far apart in longitude, it 

 can only be near noon at one place, and Error must be ob- 

 tained at the other, either on the day before or after, as well, 

 so as to be able to correct the Error to the time of interchange 

 of signals. Using land-lines, day time is most favourable for 

 exchange of signals. 



If the observation spot can be at, or close to, the telegraph observa- 

 office, it is convenient, as the watch will not have to be carried *^°^ ^P°*- 



