Seiches in Some Lakes of Japan. \ 5 



positions of the nodes remained constant; and also, wlietlier, 

 when the penduknn ball was stopped, the water continued to 

 move freely by itself. The length of the pendulum was adjusted 

 till this state was reached, and then the period was determined 

 and the positions of the nodes were noted, In short, trustworthy 

 values of the periods can only be satisfactorily obtained when the 

 water executes /ree oscillations. 



As the period of a stationary oscillation of a mass of water is 

 proportional to its length and inversely proportional to the square 

 root of its depth, we can easily deduce the period in a real lake 

 from that of the model. 



The determinations of the positions of nodes in the model lake 

 are no less difficult than the determination of the periods. Even 

 when the water is oscillating apparently in the most perfect manner, 

 the nodes shift within a small range. For the determination of 

 nodes, we may make use of either of the properties (1) that at a 

 node the vei'tical motion is zero, so that when the water is rising 

 on one side of it, it is falling on the other, or (2) that at a node the 

 horizontal motion is greatest. To utilize the first property, we 

 watched the surface of the water near a node and observed the 

 motion of the image of a distant object. The mean positions of 

 points where the image remained motionless were noted and taken 

 as the nodes. To make use of the second property, we darkened 

 the water with ink until it was quite opaque, and then we 

 scattered some aluminium powder uniformly over its surface. The 

 model was then illuminated with strong light, and the grains of the 

 aluminium powder seemed like so many bright stars on the dark 

 back ground. Now when the water was in motion, the powder 

 participated in the motion of the water, and described bright lines 

 corresponding to the amplitude of the oscillations. A photographic 



