9i Proceedings of the Royal Irish Academy. 



If we took the disturbance indicated by the first term alone of i^ in (45), 

 almost the same result would be obtained. 



The difference between these two solutions, and between their results as 

 to stability, strengthens the view that boundary-conditions are unimportant 

 if, and only if, lb is large. It is not suggested that when instability actually 

 occurs, the increase in a disturbance is as small as that obtained in the former 

 solution, or as great as that in the latter. The boundary- conditions to which 

 they refer are not those which occur in the experiment ; Ih is not large (in the 

 latter solution, very small), so that the violation of boundary-conditions is 

 important; and even the initial disturbance does not satisfy the realizable 

 boundary-condi tions . 



