method of predicting tlie thermal structure cannot be obtained 

 without consideration of the internal v:ave. 



This problem is investigated by using tx\fo models, namely, 

 the two-layer system and a model assuming the density to be a 

 continuous function of depth. 



Since the two-layer system assumes the wave to occur along 

 the interface between fluids of different densities, the problem 

 is simplified to the study oc" the oscillations of the thermocline 

 with time. It appears that the main variations in the thermocline 

 depths are: (a) seasonal variations; (b) lunar or monthly vari- 

 ations; (c) semidiurnal and diurnal tidal variations; (d) short 

 period variations; and (e) random variations due to mixing. Data 

 are presented to illustrate all these effects. 



Because previous studies have shewn (c) to be the larj^^est 

 single effect, it is studied empirically at some length. A 

 formula is presented giving the amplitudes of these oscillations 

 in one area as a function of the thicknesses of and density dis- 

 continuity between the layers by assuialng a two-layer system on 

 a rotating spherical earth. In the few cases in which the data 

 were adequate for a forecast, the relationship predicted the 

 amplitudes with a fair degree of accuracy. 



When the continuous density model is considered, a more 

 complicated picture of the internal oscillations results in 

 which the amplitudes and phases vary continuously \\rith depth. 

 Methods are developed for predicting these internal oscillations 

 and a sample prediction is presented. The observed and predicted 



