Tlie probability, therefore, of mere chance is : 



2« *^-« (360J = "■•'' 



''''^^-^ asi =''''■ 



and the final outcome 



26 days 0'.480 prob. error 0'.343 



27 ,, 0'.405 „ „ 0'.415 



If we varv the arguments given in Table VIII of Schustfr's paper 

 for a period of 26 days in such a way that the result is equi\alent 

 to an arrangement according to 25.80 davs we find : 



A,' C 



Group 



As these arguments do not differ more than 60 degrees, the probability 



of chance is in this case : 



1 



The final result, calculated for a gronp of 14 rows, 

 1M18 prob. error 0'.292. 



The accurate length of the period and I he niosl probaltJe value 

 of C, calculated after the method discussed sid) 5, are then: 

 25.804 days (' = 55.°6. 



It appears from these calculations that an arrangement accor<liug 

 to 26 and 27 days leads to results the probable incertitude of which 

 has about the same value as liie amplitude itself. On arranging 

 according to 25.8 days we find a probable incei-titude about four 

 times less than the amplitude. 



Further, from this investigation, as i-ompared with Schuster's 

 in(|uiry, we may (b'a\v the (•(uichisiou. that elements of terrestrial 

 magnetism, as observed in highci- latitudes, allow a more decided 

 juilgment to be foi-med concerning the real existence of periodical 



3 



Proceedings Royal Acad. Amsterdam. Vol. VII. 



