20 



Art. 1.— T. Terada: 



7. By the rather unexpected success of the above trial, we 

 were tempted to extend the theor}^ for the case of three-dimension- 

 al problems and put 



dp _ 



dt 



' dx' diP /' 



where x denotes the longitude and y the latitude. Putting 



= a COS 



where a is considered as the function of y only, we obtain 

 27r / \ / d'^a \^ 27r / , a; . - 



CT 



(15) 



(16) 



17) 



Hence 



CT I d'a 



27r \ dy 



y-i y 2 



47r 





if 





« , 



(18) 



If we may represent the average diurnal amplitude of temperature 

 as the function of latitude only and of the form 



a = cos^ {y - d), (19) 



where <5 is the sun's declination, we obtain 



d'^a _ ^ 5cos2(y-o")-Fl 



dy"" "" ~~ 2 



Thus «1 vanishes for 



5cos2(î/-r)) + l =0 or y-d = 50° 45'. 



In fact «1 becomes smaller at higher latitudes and shows a tendency 

 to change its sign. 



8. Returning to the case of the two dimentional problem, 



