obtained between the observed and cornputed amplitude in four 

 orders. This Is a very strong mathematical Indication of the 

 validity of his theoretical formulation. His differential equa- 

 tion, therefore, has been widely employed in studies by other 

 authors, notably Seiwell (19^2) and Lek (1938) • 



It must be emphasized that Fjeldstad was concerned prima- 

 rily with validating his theory, and not with computing ampli- 

 tudes at a time and in a place where no amplitudes were known. 

 His work has been examined in the present investigation with a 

 view to projecting his results into the future in order to pre- 

 dict the character of the semidiurnal and diurnal internal wave 

 in a given place at some future time. It was hoped that in his 

 concepts could be found soi7ie conservative or predictable factor 

 which could be employed in a technique for successful predic- 

 tion. 



The most convenient and appealing idea was that the Fourier 

 coefficients themselves were conservative with time for any one 

 place. However, no satisfactory theoretical defense for this 

 thesis could be found, since it seems clear from the above equa- 

 tions that, if the density distribution changes with time, not 

 only would the amplitude A change but also the coefficients a^. 

 This does not mean, ho^^rever, that this concept as a possible 

 basis for a prediction technique should be discarded entirely. 

 In cases of slow time rate of change of density distribution, it 

 is possible that the coefficients would roiaain reasonably con- 

 stant with time; frora a practical standpoint this v/ould be almost 

 as useful as absolute constancy. It may be that even for large 



60 



