Seiches in Some Lakes of Japan. 



11 



troublesome, if not impracticable. We found the following graphi- 

 cal method, though tentative, to he sufficiently accurate for all 

 practical purposes. It enables us to find the phase and the 

 amplitude of any component. Indeed the method assumes an 

 exact knowledge of the periods. Let Ï be the period of a motion 

 Avhose phase we wish to find. We then take from the limnogram 

 a length corresponding to 2°. T, where n is a certain integer. 

 In order to facilitate the explanation, let us take the particular case 

 of n=3, i. e. take a length corresponding to 8T. We cut this into 

 two equal parts and put one on the other, and draAV the mean curve. 



Hakoné, Sept. 2, 1901. Lak^' Hakoné. 



JO 



Pig. 6 a. 



:W" 



40" 



Fio-. 6 b. 



Fig. 6 c. 



This mean curve is again halved, and the two halves are superposed 

 and their mean is drawn again. Continuing this process, we get 

 ultimately a mean sinuous curve with a length Ï. The theory of 

 this method is that, when we consider corresponding points for all 

 these intervals of the lengtli T, the motion with the period T is in 

 the same phase, but other motions are in general in different 

 phases, so that by the process of taking the mean, the motion T 

 only survives, and all other motions destroy themselves. When 



