Computations frequently indicated significant differences between time 

 periods and/or areas. If shorter time periods or fewer areas had been 

 analyzed, it is probable that no significant differences might be found. 

 In these cases the t test was used as a detail test to compare all 

 possible combinations of pairs of means, based on the hypothesis that 

 the two means were from the same population. The results of the analysis 

 of variance and of t tests were converted into probabilities and are 

 presented in tables 11 through 20 (appendix B) . 



In these tables the column headed "X" refers to the overall mean of 

 pq., where p is the number of time periods, and q indicates the number of 

 areas having data. The reliability of the probabilities increases as pq. 

 increases. The column "monthly mean in adjacent 1-degree q.uadrangles" 

 contains the comparison of means of all data recorded during the indi- 

 cated month within adjacent 1-degree quadrangles for the particular param- 

 eter of the thermal structure. The column "monthly mean in 1-degree 

 quadrangle, year to year" refers to the comparison of means of particular 

 parameters recorded over several years for the indicated month in the 

 same 1-degree quadrangle. Therefore, table 11 may be interpreted as 

 follows: mean temperature in January, based on 9 monthly means used in 

 the analysis of variance, was 66.69°F. The probability that mean surface 

 temperatuires for the month of January in adjacent 1-degree areas are not 

 significantly different is 6o percent, while the probability that January 

 mean surface temperatixres in the same 1-degree quadrangle are not signifi- 

 cantly different from year to year is zero. In other words, there is 

 continuity between areas but not between years. 



For the remainder of the table, the conclusion can be drawn that 

 mean surface temperatures are coherent or consistent spacewise in most 

 months of the year but that they are rarely coherent from year to year , 

 except in April. During April maximum convective activity may be expect- 

 ed, and, since OWS ECHO is in the area of l8-degree water discussed by 

 Worthington (3), temperatures are consistent because of the annual water 

 mass formation process* 



Analysis of the Experimental Data 



At the various ocean weather stations, two to four BT's were taken 

 each day. The requirement of a minimum of 10 observations limited area 

 size to 1-degree quadrangles and time periods to 1 month. During the 

 four experimental surveys, BT's were taken every half hour. Thus, it was 

 possible to reduce area size to 10-minute quadrangles and time division to 

 5-day periods. 



The mean, standard error of the mean, and the standard deviation of 

 the various thermal structure parameters were computed for each of the 

 four experimental cruises and are presented in tables 21 through 2.k, 

 (appendix B). The 2-way analysis of variance and the t test also were 



