64 s. Nakannu-a and K. Honda : 



Earn, namely instead of trying to solve the equation (5) for n 

 directly, we calculate the left hand side member for some approxi- 

 mate values of n, which we know from our approximate knowledge 

 of T, and then find by interpolation that value of ii which satisfies 

 the equation (5). 



From the numbers given in table we deduce, 



h =0.38, 

 and 



7)/ = 1. 644x10^*' fl/ = ]. 644x10'" 



P, =3.728 «1=3.877 



P2 =0.848 a, =0.929 



p;=o.596 a; = 0.596 



so that 



«/ = 105.3 A =48.7 



« =248.3 ß,^2Q.ß 



«2=89.8 

 «; = 57.6 

 The result of our calculations from these data, we are obliged to 

 confess, was not at all satisfactory. While the actually observed 

 value is 1,5.38 minutes, our calculations gave 22.47 minutes. 

 The reason of this great discrepancy is rather difficult to ascertain, 

 inasmuch as Chrystal and Wedderburn found a very good 

 confirmation of the theory in the cases of Lochs Earn and Treig. 

 The reason must be sought partly at least in the fact that the 

 lakes which they had chosen, satisfy the conditions assumed in the 

 theory almost ideally, while in the present case it is quite 

 otherwise, Chrystal assumed in his theory that there is no 

 component of flow transverse to the average length of the lake. 



