96 



THE TRANSATLANTIC LONGITUDE. 



then lifted by the tension of the spring. This tension being less than the muscular 

 force of the thumb when the button was pressed down, a longer time was consumed 

 in traversino- the distance between the stops ; and, for this, repeated measurements 

 give 0\035 as a near approximation to the average intervah Now since, as already 

 related the ordinary signals record themselves upon the chronograph when the arm 

 carryino- the button leaves one stop, but are not really given until it reaches the 

 other, all the recorded intervals between the instants of giving and jeceiving make- 

 circuit signals wiU be too large by about 0'.015; while for break-circuit signals the 

 reverse obtains, and the recorded interval will be too small by about' 0'.035. Con- 

 sequently, in comparisons between break-circuit signals and others, a correction 

 must be applied, varying with the temporary adjustment of the signal-key, but 

 amounting on tire average at Valencia to not far from 0'.()5. The importance of 

 this correction will be recQgnized on inspection of tke results of the first four 

 experiments of the following series. It has, nevertheless, not been applied to any 

 of our results, inasmuch as during the exchanges between Valencia and Newfound- 

 land, no measurements or estimates were made to determine this pass-time for the 

 Newfoundland key. It must, of course, be taken into account in any attempts to 

 draw inferences regarding the relative velocity of break-circuit signals. 



The signals in these experiments were given by Mr. Mosman, and recorded by 

 myself, using the circuit formed by the two cables without any other connections 

 than the same key, galvanometer, and battery at Valencia, which had been employed 

 for the other work of the expedition. Care was of course taken that the signals 

 should be neither seen nor heard by myself, except as indicated by the deflections 

 of the galvanometer-needle. 



Exp. I. i cells. Circuit made and broken. Key between zincode and galvanometer. 



No. Mean interval. 

 Make-circuits 11 0'.25'Z 



Break-circuits 11 0.229 



Exp. II. 4 cells. The same, with 126 ohms resistance between key and galvanometer. 

 Make-circuits 10 0.279 



Break- circuits 9 0.227 



Exp. III. 4 cells. Key and galvanometer upon opposite sides of the battery. 

 Make-circuits 13 0.278 



Break-circuits 14 0.225 



Exp. IV. 4 cells. The same, with 126 ohms resistance between key and cable. 

 Make-circuits 11 0.287 



Break-circuits 11 0.220 



Exp. V. 1 cell. Positive and negative signals. 



Positive. Negative. Both. 



No. Mean. No. * Mean. No. Mean. 



2 0'.240 8 0'.292 10 0'.282 



Here the moments for the positive signals were only recognized'with difficulty, 

 8 out of 10 being lost. The battery-power was insufficient to move the needle 

 promptly, with the existing adjustment of its damping-magnet. The difference in 

 this respect between the two classes of signals was very marked, although they alter- 

 nated at the prescribed intervals of 5 and 10 seconds. 



