GG THE TRANSATLANTIC LONGITUDE. 



various delays of which our x is composed, except the actual time of transmission 

 through the cable; unless the adjustments of the two chronographic or local circuits 

 are so diverse that the loss of time which they entail cannot be regarded as equal for 

 the two instruments. This is not the case, since repeated examination has shown 

 that the difference is not measurable. The exclusive employment of signals given 

 by interrupting the galvanic circuit, and of a pen which is not removed from the 

 paper during the whole period, renders the measurement of armature-time very 

 easy, and eliminates it from ordinary observations. 



On the 2d November I made five such determinations of my personal error of 

 noting, each one based upon one series of signals, and with the following results. 

 The errors appended are the mean errors of the mean, not the so-called probable 

 errors, which would be but two-thirds as large. 



Q, 277 + o^013 



0.256 ± .012 



0.230 ± .011 



0.248 ± .014 



0.262 + 0.018 



These give the final value . . 0.253 ± O.OOG 

 A series by Mr. Mosman gave 0.275 + 0.014 

 and one by Mr. George, of the telegraphic staff, who had had no previous expe- 

 rience in observing, gave . . . 0.296 + 0.017 



On the 7th November, five determinations gave for my own error — 



0\292 ± 0^010 



0.300 ± 0.013 



0.288 ± 0.006 



0.285 + 0.007 



0.291 ± 0.010 



the mean value being .... 0.289 ± 0.005 

 Mr. Mosman's error from four determinations being — 



0.322 ± 0.027 



0.296 ± 0.031 



0.303 ± 0.016 



0.297 ± 0.013 

 and Mr. George's .... 0.309 ± 0.022 



The galvanometer was evidently somewhat less sensitively adjusted than on the 

 previous occasion, as was indeed knoAvn independently of the signals, since it had 

 been undergoing some repairs ; yet the average excess was but three and a half 

 hundredths of a second. 



The Kessels clock, at Heart's Content, was provided with two signal-giving 

 attachments, one being the ordinary arrangement for breaking circuit at the moment 

 when the pendulum-rod is vertical, and an additional tilt-hammer being available 

 for interrupting the circuit at the instant of extreme elongation on alternate seconds. 

 Mr. Dean availed himself of this means for measuring the personal error of noting 

 signals, by connecting each tilt-hammer with a separate circuit. One of these 



