Mendenhall — Free Pendulum as a Time Standard. 89 



nometer it is found to furnish a most delicate test of the 

 equality of the seconds determined by the toothed wheel 

 which breaks the circuit. 



Any inequality in the spacing of the teeth or irregularity in 

 the movement of the wheel, of the existence of which the 

 ordinary chronograph record would show no evidence, is clearly 

 and unmistakably shown, and if on account of bad workman- 

 ship such inequalities are considerable the result as viewed in 

 the telescope is ludicrous although a chronograph sheet from 

 such a chronometer might present a very good appearance. It 

 will be readily understood that when the breaks are separated 

 by uniform periods, the moving flash will advance regularly 

 and steadily, movements corresponding to one thousandth of 

 a second in time being easily perceived. When the breaks are 

 irregular, however, the flash moves irregularly, at times stand- 

 ing still for a second or two, and often actually receding. The 

 apparatus may be used therefore as a test of the regularity of 

 breaks with chronometers or clocks. 



It is often desirable to compare one chronometer with 

 another or with a clock and to obtain a relative rate. As this 

 rate is variable it is necessary to determine it by reference to as 

 short an interval of time as possible. The ordinary chrono- 

 graph record does not give satisfactory results unless the com- 

 parison extends over a number of hours or a whole day. By 

 means of the apparatus described in less than an hour a daily 

 rate fan be determined, correct within about three-hundredths 

 of a second, and a higher degree of accuracy can be secured if 

 it is found desirable. 



In this operation it is only necessary to observe coincidences 

 with both chronometers and then again in forty minutes or 

 after a longer period if a better result is wanted. No care 

 need be taken to secure standard conditions of temperature, 

 pressure, etc., as whatever changes may take place affect the 

 coincidence interval of both time-pieces alike. 



Indeed it has been found possible to get a fair rate from an 

 observation of a single pair of coincidences, the time con- 

 sumed being less than five minutes. It is evident that a 

 method is here offered for studying the hourly variations of 

 chronometers and clocks, due to temperature, pressure, or 

 other changes and in making such investigations the time- 

 piece under examination finds its severest critic not in another 

 of its own kind, but in a free pendulum maintained under 

 uniform standard conditions. 



As an example of the comparison of two chronometers by 

 this method the following is cited : 



