1916] on Illusions of the Upper Air G15 



which it offers to the steady progress of air. The air loses some 

 of its motion, and is exposed to the pressm-e without the velocity 

 that is required to balance it. It must, therefore, fall away towards 

 the low pressure, taking out of the pressure the energy necessary to 

 provide for the loss by friction. Thus the convergence which we 

 have to account for is only that shown near the surface within half a 

 kilometre. We need not trouble ourselves about a supposed con- 

 vergence and convection over the whole area in the upper air. The 

 second element of our specification disappears. After years of con- 

 templation of the motion of the air from high to low as produced 

 in a quiescent atmosphere by the operation of pressure-difference and 

 kept within bounds by friction, we now regard the motion from high 

 to low as actually caused by the friction which retards the velocity 

 required to maintain the strophic balance. To base the theory of 

 motion of the upper air upon the idea of a given distribution of 

 pressure setting a quiescent atmosphere in motion is as great an error 

 as to begin the lunar theory by supposing the moon to start from 

 rest under the force of the earth's attraction, and only to find out 

 after it had started that the earth was moving. 



As to convection, there is certainly convection wherever there is 

 the juxtaposition of air of different densities ; it takes a great variety 

 of forms, it is very common in cyclones, but it is not a necessary 

 attribute of them. Possibly it is set up there more easily because 

 the air travels so much faster in cyclonic areas than it does in anti- 

 cyclones, and adjoining localities are fed from different sources of 

 supply. The convection is probably the one disturbing cause of the 

 strophic Ijalance of velocity and pressure. So we regard the tropo- 

 sphere as a layer of about 9 kilometres thick, always striving to 

 arrange its motion according to the pressure, and perpetually baffled 

 in its endeavours by the ubiquity of convection. Since all the 

 changes proceed by infinitesimal steps, there is never a time when 

 we can identify a state of finite divergence from the balance between 

 velocity and pressure. From this point of view the centre of a 

 cyclonic or anticy clonic system has no special dynamical importance. 

 It becomes a notable feature on the map, when for any reason the 

 cyclostrophic component is the chief element in balancing the pressure. 

 That is seldom the case in our maps, which more often consist of 

 isobars of complicated shapes. 



The Dominance of the Stratosphere. 



Further than this, Mr. Dines has thrown a new light upon the 

 origin of differences of pressure at the surface by obtaining the 

 correlation co-efficient between corresponding deviations of pressure 

 from the normal at the level of 9 kilometres and at the ground, and 

 has obtained results ranging from 0.G7, for the last available set of 



