1589. 







1518. 





m. 



s. 



h. 



m. 



s. 



17 



15 



21 



07 



10 



14 



21 



22 



03 



48 



57 



06 





56 



38 



90 Men den hall — Free Pendidum as a Time Standard. 



Chronometers. Negus. 



h. 



1st coincidence 18 



2d coincidence 19 



Ten coincidence interval.. 



No. 1589-No. 1518 = 28 seconds. 



From which the rate is found to be one second per day. 



If the claims here made for the free pendulum are justified 

 it is likely to find its most useful application as an independent 

 time standard. For this purpose not less than three of these 

 pendulums should be used. They should have slightly differ- 

 ent oscillation periods and these must be determined with 

 great care, under standard conditions and referred to the most 

 accurate mean time second available. Indeed for the first 

 standard set it might be worth while to carry on comparisons 

 during day and night for many days, that all errors incident 

 to the determination of time may be eliminated as far as pos- 

 sible. Assuming the physical constancy of these pendulums 

 and the possibility of reproducing with sufficient accuracy the 

 conditions under which they were swung, no further reference 

 to astronomical observations is necessary, if, only the force of 

 gravity does not change. Such a set, or one derived from it 

 may be issued to any point, with a certificate guaranteeing the 

 periods of oscillation as a simple function of the force of 

 gravity. If the latter is known the oscillation period is at 

 once deduced. Whenever it is desired to rate a clock or 

 chronometer it will only be necessary to observe a series of 

 coincidences, using the pendulum and chronometer as de- 

 scribed above. The use of three pendulums, whose differences 

 must remain sensibly constant wherever they are, will serve as 

 a check against changes of any kind and once in a few years a 

 redetermination by means of astronomical time may be had as 

 a further security against systematic error. The ease and 

 accuracy with which their oscillation periods can be determined 

 in terms of the second, of a clock or chronometer, and the 

 probable constancy of this period during long intervals of 

 time encourages the beljef that the issue of properly authenti- 

 cated time standards is by no means an impossibility and as to 

 the desirability of an early accomplishment of such an end 

 there can certainly be no doubt. 



