applied to each 5-day-lO-niinute oLuadrangle sort. The resulting probabil- 

 ities are presented in tables 11 through 20. The 5-day means of adjacent 

 lO-minute quadrangles and alternate lO-minute quadrangles are results of 

 testing means of all data recorded within adjacent and alternate 10-minute 

 quadrangles during the same 5-day period. The successive 5-day and alter- 

 nate 5-day means are results of testing means of all data recorded during 

 successive and alternate 5-day time periods within the same 10-minute quad- 

 rangle. Blank spaces in the tables indicate lack of data. 



If the data are assvuned to have been taken at random over the 10-min- 

 ute quadrangles dioring the same 5-day periods, the tests of adjacent and 

 alternate 10-minute quadrangles indicate significant difference in the 

 indicated thermal structure parameter over approximately 10- and 20-mile 

 distances. 



Referring to entries in table 11 reflecting resiolts of experimental 

 observations (columns involving 5-day means and 10-minute quadrangles), 

 there is good probability of no significant difference between adjacent 

 areas for all four experiments, whereas there is probably a significant 

 difference between 5-day means for the experiments in March, May, and 

 November, indicating significant short-term thermal changes which affect- 

 ed all areas more or less simultaneously. These short-term changes are 

 of broad scope affecting major portions of the ocean. 



RESULTS OF ANALYSIS 



Appendix C contains graphs of the population mean monthly values of 

 each parameter of the thermal structure (historical data from the 20-minute 

 quadrangle centered at 35°N,48'-V). 



Experimental data were used as a reliability check of the historical 

 data. If the means of the experimental data fell within the range of the 

 historical maximum and minimum means, the historical data were considered 

 to be reliable. Means for March generally were not valid because of the 

 lack of data, or because comparisons indicated that the historical data 

 were in error. Comparisons also indicated that the mean gradient is 

 dependent on the type of BT employed (I150 or 900 feet). Thus, parameters 

 of the mean gradient of the historical data are not accurate. All compar- 

 isons can be observed in the graphs in appendix C. They are also listed 

 in column 5 of table 25 (appendix B) . 



When the population mean monthly values were plotted the resulting 

 circves were sinusoidal. If the comparison of the means (historical ver- 

 sus experimental data) indicated reliable historical data for all months, 

 the parameter curve was smoothed by Fourier analysis on the I/JP-SO com- 

 puter. 



The resiilts of the Fourier fitting are: 



Surface Temperature (2) 



y=71.60- 6.61 sin x - 2.68 cos x + 1.29 sin 2x + 0.36 cos 2x 



