ON SEISMOLOGICAL INVESTIGATIONS. 



65 



Hence the factors are as below : — 



Days 2 3 4 5 



Factors 8 12 15 17 



An example may make the method clearer. The actual readings of 

 the Seychelles records are given in columns 2, 3, and 4 of Table VI. 

 The unii is 01 in. ' 



Table VI. 



In the next two columns the simple differences between July 12 and 

 15, and between July 15 and 18, are shown. These are clearly affected 

 by diurnal terms, i.e., the diurnal terms differ for different days, as we 

 might expect. To bring out the lunar terms more clearly we remove 

 the diurnal terms, including a suitable constant. The terms found 

 (by harmonic analysis) are shown in the next two columns, and finally 

 the corrected differences, which show the semi-diurnal terms clearly. 

 Analysing these harmonically (as we could of course have done without 

 removing the diurnal terms), we find : — 



For July 15- July 12 : 



in. in. h. 



0-048 cos 2 (^-28°) =0-048 cos 2 (< - 11-9). 



For July 18- July 15 : 



0-045 cos 2 (fl-52°) = 0-045 cos 2 {t - 13.5), 



where 6 is the hour angle measured from 10 h., or t is the time in hours 

 measured from h. The hour of maximum has thus advanced 1-6 h. 

 in the 3 days. A purely lunar tide would advance 2-4 h. in 3 days. 

 The discrepanc}' is partly accidental, partly due to a semi-diurnal tem- 

 perature effect, which can only be detected or eliminated by a longer 

 series of observations. T3ut we can clearly separate the lunar effect by 

 its advancing phase if we have a long enough series of days. 



We proceed to give a few results for stations which had sent films to 

 Shide for examination. 



The measures were made in hundredths of an inch, and the travel 

 of each trace is about 0-23 in. per 2 hours. The figures below, being 

 deduced from only a few days' records, must not be taken too seriously, 



1915. p 



