762 



DR E. M. WEDDERBURN AND MR A. W. YOUNG ON 



half of the observations, when the amplitude was considerably greater than during 

 the second half. The maximum period as far as the periodogram of these observa- 

 tions indicates is not much different : but from the point of view of the periodogram 

 itself, it will be noticed that the use of only half the number of observations has 

 lessened the " steepness " of the curve to a very great extent. 



Periodogram Analysis. 



If Isothermal. 19/3 August 24-29. 



Fig. 14. 











PERIODOGRAM: 











Showing Maxima 



AT 10 HOURS 









and 



AT 10 HOURS. 









<r^S *"^ 



'7 ^^ 



10 



15 



20 



25 



Mean 20-hour Oscillation.- 



— Uninodal Seiche. 



Mean 10-hour 



Oscillation.- 

 — Binodal Seiche. 



Fig. 15. 



Fig. 16. 



The average 20-hour (taken for simplicity as the maximum instead of 19 '5 hours) 

 and 10-hour oscillations are shown in figs. 15 and 16. It will be seen that the 

 10-hour curve is a fair approximation to a sine curve, but that the 20-hour curve 

 is not. The reason for this is that the periodogram method is unable to separate 

 the 10-hour oscillation from one that is so nearly of double the period, so that the 

 average 20-hour oscillation as given by the periodogram is really a combination of 

 the uninodal and binodal components. 



