Seiches in Some Lakes of Japan. 9 



holes was bored, through which a thm thread was passed, as is 

 shown in the accompanying figure, so that we had a number of 

 thread lines which were parallel to one another and the distance 

 between them could be varied at will by deforming the frame work. 

 This frame was put on the limnogram curve in question, and the 

 distance between the lines was so adjusted that the threads best 

 coincided with the maxima and the minima of the sinuous curve. 

 Then by measuring the distance between the threads we could 

 easily deduce the corresponding period from the known rate of the 

 clockwork of the recording cylinder. 



Another method was of great help in determining the periods 

 accurately. We shall call this " the method of coincidence." It 

 was simply this. It often happened that a given condition of 

 complex oscillation continued unchanged so that the amplitudes of 

 the component motions did not diminish for a long time, and 

 therefore when we put one limnogram upon another and moved 

 them suitably, we could bring them to coincide closely. Now when 

 the coincidence is perfect, there must be certain integral numbers of 

 component waves during the time, which elapsed between the two 

 curves. These integral numbers can be found from our previous 

 knowledge of the periods. Dividing the time interval above found 

 by these integers we get more reliable values than before. It is 

 needless to say that it is better when possible, to make this interval 

 as long as can be done reasonably. We say reasonably, for Avhen the 

 interval is too long, there will be some uncertainty as to the integer 

 serving as the divisor. The curves Figs. 5, a and h exemplify the 

 method of coincidence. By placing the curves one upon the other 

 and displacing them horizontally, they will be found to coincide 

 pretty well when the points in equal phases are separated by an 

 interval of 199.8 minutes. Dividing this by 15.4, 6.7, and 4.6, the 



