TABLE XXV. 



Calculating the probable errors from Table XXV, we get for the probable error 

 of the difference 



M p -7 

 21 M,. 



4-9 x 10 7 sec., comparing with e- = = 12-7X10' 7 sec. 

 13-0 e. --= 24-6 



14-3 e a =ll-9 



mean value 



10-7 X 10~ 7 sec. mean value 



1G-4X10- 7 sec. 



The mean value 10-7 x 10~ 7 sec. compares with 3- 1 x 10~ 7 sec. for the first Potsdam 

 observations, and 2-5xlO~ 7 sec. for the second. 



This measure is 3-4 times as great as that for the Potsdam observations and suggests 

 that the coincidence apparatus was probably the cause of no small part of the large 

 probable error. At the same time, it must be remembered that a source of error which 

 is included in this estimate, is that due to rariulioim chiefly in temperature and clock 

 rate during the progress of a single set. These may have been considerable in the 

 series. 



39 



c 4 



