days, the differences are not randomly distributed. As examples, 

 five such time periods are noted on figure 12. It is noted that 

 period four persisted for only a few days while period five appears 

 to have persisted for a month or more. It is concluded that equa- 

 tion (4) does estimate the seasonal variation, but does not, as 

 expected, estimate the shorter-period temperature variations 

 related to short -period transient phenomena. 



Figure 13 shows the differences as a function of 10-minute- 

 latitude intervals. These differences appear to be randomly 

 distributed about zero difference. The numbered short -time- 

 period samples shown on figure 12 are also indicated on this 

 figure, giving an indication as to the latitudinal extent of these 

 short -period anomalies i 



Figure 14 shows the differences as a function of 10-minute 

 longitudinal intervals. Again the differences appear to be ran- 

 domly distributed about zero difference. Also the numbered short- 

 period samples are shown, giving an indication of the longitudinal 



54 



