



Number of 





100 r2, 



(J, 



Year 



Longitude 



Observationp 



R 



percent 



degrees F 



SEPTEMBER 











1956 



126°W 



37 



0.88 



76.7 



2.6 





132° 



44 



0.90 



81.9 



2.3 





138° 



41 



0.95 



89.7 



2.1 



1957 



126°W 



41 



0.84 



71.1 



2.1 





132° 



59 



0.88 



78.4 



2.3 





138° 



52 



0.96 



92. 



1.7 



Relatively the fit is best for data taken the greatest distance 

 from shore, the percent variance explained by regression varying 

 systematically from 94 to 71 percent. The reader is reminded 

 that the original data were reported to the nearest whole degree 

 Fahrenheit and that many of the temperatures are "injection" 

 temperatures taken at some depth below the surface. The standard 

 deviation would be expected to be greater for these data than for 

 the data used to establish the seasonal variation. 



On figure 2 the solid line is a plot of equation (2) using the 

 regression coefficients for the proper latitude, year, and month. 

 A third-degree polynomial appears to exhibit the flexibility neces- 

 sary to obtain a reasonable estimation of the latitudinal variation. 



It is of interest to note that from 30 to 40 N the temperature 

 decreases about 1. 2 F per 1-degree change in latitude, about 60 

 nautical miles. 



27 



