47 



beginning of each second, and if we omit the slight notch 

 which would be made by the fifty-ninth second we can 

 thus register the beginning of each minute. This tele- 

 graph key which I hold in my hand is in the same electric 

 circuit with the clock and chronograph, and as this star is 

 passing over the screen I can register its transits upon the 

 chronograph by simply causing the pen to make a notch 

 in the line by breaking the electric circuit. We have put 

 a telegraphic sounder in the same circuit, so that I think 

 you will be able to hear the beats of the clock quite to the 

 other end of the hall. There comes our star, and as it 

 crosses each wire your will hear the familiar telegraph 

 tick which tells us that we have made the slight notch on 

 the chronograph which records the star's transit. Now 

 suppose that this slight notch we afterwards find occurs 

 six-tenths of the way between the thirtieth and thirty-first 

 second. Then we know that the star's transit occurred at 

 thirty-one seconds and six-tenths of a second of a particu- 

 lar minute. After an evening's observation the sheet is 

 removed from the cylinder, labelled, and filed away with 

 the records of the Observatory. 



Having obtained the error of our time-piece to within 

 a twentieth of a single second, the next question is, How 

 shall we keep the time-piece so that it will have the same 

 error to-morrow night, that it has to-night; or failing in 

 this, how shall we preserve the same relation between the 

 errors on consecutive nights? In other words, how shall 

 we know that the time-piece gains or loses regularly? 

 This leads us to speak of the clocks and watches which 

 keep the time from day to day. You all know that the 

 test of the performance of any time-piece is found in 

 noting the regularity with which it gains or loses. Thus 

 a clock which gained ten seconds a day might be a very 

 much better one than another which gained and lost alter- 



