THE TRANSATLANTIC LONGITUDE. 65 



the galvanometer signals. To obviate this difficulty, a battery of two Minotti cells 

 being employed, the circuit was divided at the galvanometer into two branches — 

 one, of fine German silver wire, passing to the galvanometer and thence again to 

 the main circuit, Avhile the other branch was made to pass through the break-cir- 

 cuit key by means of which the deflections were noted. The resistances of these 

 two branches were so adjusted that they were in the ratio of 1 to 100, by which 

 device each signal at the observatory was sharply indicated on the galvanometer, 

 without too great violence ; and by a slight adjustment of the movable permanent 

 magnets, it was always possible to render these deflections similar in amount to 

 those received from Newfoundland. It Avas, of course, necessary to include the 

 clock in the galvanic circuit, in order to obtain a time-scale ; but the interruption 

 and restoration of the circuit at each oscillation of the pendulum caused a vibration 

 in the galvanometer needle, which was not quieted for more than half a second, 

 and then only to be renewed immediately. To render all the circumstances of the 

 experiment as similar to those of the longitude-signals as the nature of the case 

 permitted, as well as to avoid any tendency to mechanical rhythm in the act of 

 noting the signals (a source of inaccuracy which every observer by the chrono- 

 graphic method must have recognized whenever the beats of his clock have been 

 audible or visible during the process of observation), it was necessary to dispense 

 with the clock while the measures were actually in process. 



The observations were therefore arranged as follows: — After the clock had been 

 included in the circuit for some minutes, recording its beats upon the chronograph 

 in the observatory, and manifesting them likewise upon the galvanometer in the 

 telegraph office, the assistant in the observatory excluded the clock from the circuit 

 by means of a plug-switch, thus stopping all record of time upon the chronograph 

 sheet, although the pen continued to trace a straight line, and stopping likewise 

 the pulsations of the galvanometer needle, by which indication the observer was 

 warned that the signals were about to begin. He then gave a set of ten signals on 

 one of the observing keys, at the same intervals, roughly, as those exchanged for 

 longitude — namely, four sharp, quick taps upon the key, about five seconds apart; 

 then, after ten seconds, three more ; and, after another ten seconds, yet three more. 

 At the close of this set of signals, he restored the clock to the circuit by removing 

 the plug from the svi^itch, and the graduation of the time-scale recommenced as 

 before after an intermission of scarcely a minute ; so that the times of each signal 

 could be read off by means of the second-marks of the preceding and following 

 minutes with an accuracy scarcely, if at all, inferior to that attainable when the 

 time-record is simultaneously in progress. The chronographic records of the signals 

 thus given are about 0'.04 long. 



The observer is meanwhile at the galvanometer in the other building, out of 

 sight and hearing of the assistant, and notes the moments of deflection of the 

 light-spot by a tap upon the break-circuit key which he holds in his hand, taking 

 care to conform in all respects to his habitudes while observing longitude-signals. 

 The intervals between the chronographic records of the original signals and his 

 observations of the same, then furnish a measure of the " personal error of noting" 

 as already defined ; and show the lapse of time corresponding to the sum of all the 



