I. INTRODUCTION 



In June and July 1953 Operation STANDSTILL was conducted in the area 

 northeast of Bermuda. During the course of this operation an anchor station 

 was maintained at 33°33 ! N, 62°25'W for a period of 25 days, from 11 June to 

 6 July e Bathythermograph readings were taken every half hour as part of the 

 oceanographic observations made at the anchor station. The following dis- 

 cussion deals with one aspect of these observations r the analysis of the 

 wavo spectrum of internal waves. 



II. INTERNAL WAVES 



Previous discussion of the theory and existence of internal waves in 

 the ocean has been directed toward the demonstration of the occurrence of 

 internal waves of tidal period. Thus, Haurwitz (1953) has examined the 

 temperature data from Meteor stations 385 and U38 and the Altair anchor 

 stations and has concluded that the oscillations at Meteor station 385 

 show possible existence of internal waves of semidiurnal tidal period. 



With the recent development of the techniques of power spectrum 

 analysis (Pierson, 1952), it is possible to analyze the spectral components 

 of any suitable series of data comprising a Gaussian process. If it is 

 assumed that the fluctuations of temperature in the ocean represent a 

 stationary Gaussian process, the power spectrum analysis will indicate 

 the significant periods present in the series and the power associated with 

 these periods. In this way, it is possible to determine whether internal 

 waves of tidal periods exist, as well as other periods greater and smaller. 



III. ANALYSIS OF DATA 



An examination of all the BT data showed that the deepest continuous 

 isotherm was 66° F. Accordingly, the depth of this isotherm was determined 

 from each half -hourly bathythermogram. These data are presented in table 

 1. Two ships in succession participated in this operation, the USS RE- 

 HOBOTH and USS SAN PABLO. The frequency of the readings of the depth of 

 the 66° F. isotherm is given for 25-foot intervals from 76 to 1,100 feet. 

 This frequency distribution is graphed in figure 1. 



Theoretically, distributions which are to be analyzed by power spectrum 

 methods should be normal, i.e., Gaussian (Tukey, 19il9). The present dis- 

 tribution has an upper limit at the surface and therefore is of the Pearson 

 Type III. A normal distribution curve was fitted to the observed distri- 

 bution, taking only the depths from 7 6 feet to 575 feet. The resulting 

 curve is shown in figure 1 as a dashed line. The fitted curve indicates 

 that the observed curve does not depart greatly from the normal distribu- 

 tion. Thus, since the frequency distribution is nearly normal, power 

 spectrum analysis by the usual methods will not give meaningless results . 

 The series of 1191 readings was analyzed for its power spectrum (Fierson 

 and Marks, 1952) in five different ways! 



