350 SCIENCE PROGRESS 



gradient-wind cannot be determined immediately at a single 

 station ; it is nevertheless desirable to keep the idea as the 

 basis of a standard of reference for wind measurement. 



Let us next consider the general structure of the atmosphere 

 as disclosed by observations of pilot balloons. The process of 

 observation consists in noting from minute to minute the posi- 

 tion (altitude and azimuth) of a small balloon, inflated with 

 hydrogen, by means of a theodolite or two theodolites and 

 computing the vertical and horizontal distances traversed in 

 each minute. We cannot at present make any general state- 

 ments as to the structure of the upper air from observations of 

 this kind ; we can only give some examples and say that the 

 cases which have been investigated may be grouped in 

 certain types. In the diagrams (3) of fig. 2 the wind direction 

 and the wind velocity are represented separately. The 

 gradient velocity is marked on each of the diagrams. On 

 examination it will be noticed that all types have a common 

 characteristic, namely, that the wind velocity increases from the 

 surface and approaches the gradient velocity with a veering of 

 the direction of the wind through about two points. The height 

 at which the gradient velocity is reached is not uniform but, 

 generally speaking, a near approach to it is made by the time 

 the level of half a kilometre is reached. Therefore we may 

 divide the upper air into two sections, the first of which reaches 

 from the surface up to the level of the gradient velocity of 

 surface isobars, the second extending from the lowest level of 

 the gradient velocity upwards. 



We have a great number of observations of the velocity of 

 the winds in the first 3,000 ft. from soundings of the air by 

 means of kites as well as by pilot balloons (4). The results on 

 different occasions are somewhat irregular but one conclusion 

 may be briefly stated, namely, that the change of velocity with 

 height is more fairly represented by increases proportional to the 

 surface velocity than by the addition of fixed increases depend- 

 ing on the height and not depending upon the surface velocity. 



We get the upper wind from the surface wind by multiply- 

 ing by a factor, not by successive additions of a fixed sum. 

 From this generalisation we have the glimmering of a law that 

 the velocity of the wind is proportional to the surface velocity 

 and to the height above a datum level. The datum level is not 

 perhaps really the same for all ascents, all wind directions and 



