( 25 ) 



\lll1lI)(M' of" 



1—17 1.66 289' 510 



From this it a[)|)ears that there is certaiiilv sdiiie ijiclicafioii for 

 the existcnoe of a periodieal oscinalioii, and also that the arrange- 

 ment lias been made aecordiiig to a period which practically leads 

 to a maximum value of the amplitude. 



The probability that three poiuts, taken successively at I'andom, 



are situated within an angular space of 80" is f y- J and the probabilily 



of mere chance would have been even less if we had taken into 

 account that the amplitudes too are i]i good accordance. 



5. A second, ecpially sinijtle method is allbrded I)y a direct \ie\\ 

 of the outcome of the arrangement itself, s|)lit up into two or nioi-e 

 grouj)s. 



Fig. 1 gives a grai)hical represenlaliou of llie dilfei'ences given in 

 the three last columns of Table II. 



Fig. 1 sllo^vs ilial ihe c^lr^•es of ihe Iwo series agree satistacloril_\ 

 and also that a lendency to a double period, with a maxiiiinni 

 on the 8 — 9^'' day. which in Ihe lirst urouji is still well marked, 

 vanishes when the arrangement is continued. 



If these results are considered as fairly conclusixe, so as to justify 

 a more exact determination of the length of the period, this may 

 be easily done by \ai'ying the ai'guments ^ ' of Table 1 successively 

 by ^L'l', ^/.2-i\ '',.,■'' t'tc., r denoting iIk^ \arialion of each group- 

 argument which leads to the most constant \alue of (\ 



In this way 17 ecpiations are obtained from which the luosi 

 pro[)al)le values of ^'and the period 7' can be calculaled. If to each 

 e<piation the weight is given of the coi-responiling amplitude, the 

 equations will assume the foini : 



0.69 (- 118" + '/._, ,/■) = 0.69 C 

 7.63 ( — 73" 4- 7, X) = 7.63 C etc. 



