1916] on Illusions of the Upper Air 621 



of pressure produced in the troposphere is unimportant. The distri- 

 bution of the stratosphere is dominant throughout the troposphere. 



3. The apparently capricious variations of wind and temperature 

 with height disclosed in pilot-balloon ascents and hallons sondes. 



The results of the observations of hallons so?ides show local varia- 

 tions of temperature, and those of the observations of pilot balloons 

 show similar variations of the direction and velocity of wind. These 

 variations can be connected numericallv by Equation A in combina- 

 tion with Equation 1. A number of examples are given in a paper 

 read before the Royal Meteorological Society. To quote one, the 

 rapid transition from a southerly wind at 1100 metres through a 

 calm to a northerly wind at 1500 metres on October 16, 1913, was 

 shown to indicate a temperature-gradient of 7° per hundred kilometres 

 towards the east, a condition that was in satisfactory accord with the 

 meteorological circumstances of the time. 



The same combination of equations enables us to specify the con- 

 ditions under ivhich ivhat is Jmoirn as " EgnelVs laiv " — that wind 

 velocity at different heights is inversely proportional to the density at 

 those heights — may be expected to be verified, and the conditions 

 prescribed are essentially reasonable. 



4. The rapid fcdling off of wind in the stratosphere noted i7i observa- 

 tions with pilot balloons. 



This is illustrated by a diagram compiled from the figures of high 

 soundings reproduced in Captain Cave's " Structure of the Atmo- 

 sphere in Clear Weather " (Fig. 5). The result follows directly from 

 the appHcation of Equation B to the special conditions of the strato- 

 sphere. The computations for the four occasions in which there was 

 a wind of considerable magnitude at the base of the stratosphere 

 give the following results : — 



The calculation has been arranged to give the computed horizontal 

 temperature gradient, because the values of that quantity are indicated 

 in the models of temperature distribution constructed in the Meteoro- 



