68 THE TRANSATLANTIC LONGITUDE 



For the difference of the time-scales 52 comparisons, during one minute pre- 

 ceding the observations, give 0'.491, and 60 comparisons immediately afterwards 

 give 0'.499. Adopting 0\495, therefore, as the most probable value, and deducting 

 this from the final value 0'.830, we have 0'.335 as Mr. Dean's personal error in 

 noting the signals. 



The difference between this error and that found for my own observations at 

 Valencia is small, and is probably owing to the galvanometer rather than the ob- 

 server ; the apparatus at Heart's Content being known to be somewhat less sensi- 

 tive than that at Foilhommerum. The constancy of the error is also here strongly 

 manifest ; and the illustration of the unrecognized but marked effect of the sound 

 of the tap, upon observations supposed to be of the visible deflection only, is in- 

 structive. 



It may not be inappropriate to mention in this connection that a very marked 

 effect upon the observation of transits of stars is likely to be produced when the 

 chronograph is in the same apartment, so that the regular beats of the magnet are 

 audible. When the intervals between the transit-threads are approximately multi- 

 ples of half a second, the tendency is very great so to tap upon the observing key as 

 to produce a rhythmical beat in the armature ; and when the interval differs from 

 the multiple of a second, the occurrence of that magnet-beat which records an even 

 second often precipitates the tap of the observer, whose nerves are in keen tension 

 awaiting the instant of bisection. Only a strong effort of will can obviate these 

 perturbing influences — which are akin to those exhibited in the measurements just 

 described. 



The personal error of noting being then assumed as 0\21l at Valencia, and 

 0".335 at Newfoundland, the sum of these quantities, or 0\606, is to be dedvicted 

 from our value of Xi + x., to obtain the true time of transmission ; and half their- 

 difference, or 0'.032, is to be deducted from the longitude after all other corrections 

 are applied. This correction will be taken into account, in fixing the value to be 

 adopted. 



It may be added that the indications are strong that a considerable portion of 

 this " personal error of noting" is not strictly a personal phenomenon, but that it 

 is due to the consumption of a very appreciable interval of time in overcoming the 

 inertia of the needle and in moving the needle through an arc sufficient to attract 

 attention. Indeed it is my conviction that not less that the tenth of a second is 

 thus lost. 



An automatic apparatus might be arranged, all other means failing, for recording 

 the signals received, by adjustment of delicate silver wii'es on each side of the galva- 

 nometer needle, in such a position, and so connected with the battery, that they 

 would be brought in contact whenever the deflection of the needle reached a cer- 

 tain angle, and the signal be thus recorded upon the chronograph. This would 

 definitely decide the question ; but, for obvious reasons, no such experiment was 

 undertaken at Valencia. My immediate object was thoroughly attained by the J 

 satisfactory results of these measurements of the sum of all delays not due to time 

 consumed in the actual transit of the signals across the Atlantic. 



