154 Proceedings of Royal Society of Edinburgh. [sess. 
If a vertical transverse barrier be introduced into the liquid 
near a node, it will prevent the natural movement of the particles 
there, and thus interfere with the free oscillations of the liquid ; 
but if such a barrier be interposed at a ventral segment, it will not 
interfere with the motion of the particles which are there moving 
vertically. In this connection evidently, the uninodal period of 
each division, in the case of a barrier inserted at the middle of a 
symmetric parabolic curve, is the binodal period of the complete 
parabola. 
The following observations made during the investigations 
above detailed, though of a secondary nature as regards the 
primary object of this paper, are submitted as being perhaps of 
some intrinsic significance. 
(1) In most cases no difficulty was experienced in generating 
the various seiches beyond that of ascertaining by repeated trials 
the period of the appropriate impulses. But in the particular case 
of the convex parabola it was found practically impossible to 
create the uninodal seiche by any other method than by forcing a 
vertical motion on a portion of the water, separated from the main 
body by a partition, on the same principle as the spiral spring 
mentioned above. 
Even by this means it was difficult to get a pure seiche. As a 
rule the motion would continue apparently regular for two or 
three oscillations, but would very quickly become somewhat erratic 
/ 
r 7. 
/ \ 
F Float. 
S Pointer (straw). 
