THE TRANSATLANTIC LONGITUDE. 67 



passed through the normal signal-apparatus of the clock, the Morse register, the 

 signal-key, and the galvanometer ; the other through the subsidary tilt-hammer, the 

 observing key, and the chronograph. The original signals were thus recorded on the 

 chronograph sheets by means of a clock-scale graduated to two seconds, while the 

 observations of the same were registered upon the Morse fillet ; and a slight change 

 made in the connections at the close of the experiments sufl[iced to put the records of 

 both tilt-hammers upon the chronograph, and thus permit an accurate measurement 

 of the interval between the two systems of clock-signals. In the first-named circuit a 

 battery of two carbon cells was employed, resistance-coils being interposed to reduce 

 the deflections of the galvanometer to the magnitude of those obtained through the 

 cable ; and the chronograph magnet proved sufiiciently sensitive to record these. 

 On November 10, five series of measures gave for his personal error — 



0^22 ±0'.020 





0.28 ± .019 





0.24 ± .010 





0.24 ± .025 





0.22 ± .008 



or from all ... 



.... 0.236 ± 0.009 



On the 12th November, again, his observations of twelve sets of signals give, 

 after deducting 0'.48 from each to correct for the difference of the two time- 

 scales— 



0.224 ± 0^017 







0M48 + 0\010 



.244 ± .020 







.195 ± .009 



.193+ .014 







.208 ± .016 



.181 + .014 







.271 ± .012 



.170 ± .017 







.239 + .022 



0.239+ 0.016 







0.209 + 0.013 





. 0M92 ± 0' 



.009 





The marked inferiority of these values to those found for three observers, on two 

 different occasions at Valencia, excited my suspicions, and on mquiry of Mr. Dean 

 it proved that his observations had been made in the same room in which Mr. 

 Goodfellow had given the signals, and where the click of the key was distinctly 

 audible, so that the observation was not purely dependent upon the deflection 

 of the needle, but was possibly influenced by the sense of hearing. 



Mr. Dean therefore repeated his observations under circumstances precluding 

 the possibility of his personal error being affected by any extraneous influence of 

 this kind. This was done on November 17, and ten series of signals (one of the 

 original eleven being discarded for manifest irregularity) afford the following 

 results, in which the difference of time-scales is included : — 



0\803±0\027 0^832±0^020 



.834 



.014 



.831 



.020 



.820 



.017 



0.848 



0.021 



.795 



.018 



.857 



.026 



.870 



.027 



0.864 



0.026 



the definite value from the ten series being 



.830 ± 0\008 



