1904] 



Circulation of the AtmospJwre. 27 



easterly flow of air on the polar side. It is very irregularly marked 

 in the northern hemisphere, owing to the distance of the pole of cold 

 from the geographical pole ; in the southern hemisphere, the arrange- 

 ment is probably much more symmetrical, owing to the coincidence of 

 the pole of cold with the geographical pole. 



The line of minimum average pressure forms the storm track of the 

 circular storms, resulting from the merging of the two circulations. 

 It is evident that such merging must take place. If circumstances 

 were so adjusted that there was an equilibrium condition along any 

 level, the difference of density of air at different temperatures would 

 cause a departure from equilibrium conditions, in one direction above 

 the prescribed level, and in the opposite direction below it. And to 

 this primary cause of disturbance of the steady motions corresponding 

 to the two separate distributions of pressure, must be added the 

 instability, which is due to the condensation of water vapour. The 

 region of minimum pressure becomes, therefore, the scene of great 

 changes of energy, cyclonic circulations, and variable winds. It may, 

 indeed, be noticed that increase of entropy when mixing takes place, 



is represented by _Lz__/ (supposing T-T to be small com- 

 .J JL JL 



pared with T or T'), where H is the quantity of heat transferred in 

 mixing, and T and T' are the absolute temperatures of the amounts of 

 air, supposed equal, which mix. Hence the entropy change is greatest 

 where there is the greatest temperature difference, but, for the same 

 temperature difference, where temperatures are lowest. 



To sum up the result that has been arrived at, the average dis- 

 tribution of pressure can be divided into two parts, neither of which 

 shows the characteristics of local cyclonic distribution ; one part, due to 

 the upper atmosphere, favours a westerly circulation, the other, due 

 to the lower atmosphere, favours an easterly circulation, and both 

 circulations follow roughly the lines of the average isotherms. 



This result leads to many suggestions of considerable importance. 

 The second component of the distribution is obviously directly 

 dependent upon surface temperature, and must be changed when 

 surface temperature changes; the first may also be regarded as 

 depending ultimately upon surface temperature, for its lines follow 

 those of surface temperature on the average, but the connection must 

 be less direct. It may possibly indicate differences of thickness of the 

 atmospheric layer as distinguished from differences of density in a 

 layer of uniform thickness. The first effect of a change in the dis- 

 tribution of surface temperature will be to change the character of the 

 second component, leaving the first component unchanged, except in so 

 far as expansion of air in the lower strata alters the pressure at a given 

 level in the upper strata. The study of the effects of recognised 

 changes of temperature distribution upon the second component, which 



