486 On an Integrating Anemometer. 



between them, so that an increase of n on the counter denotes 

 kn revolutions of B. 



Letting §, as before, be the length of the bar F between its 

 points E (fig. 2), 8 is the greatest distance which the wheel 

 W can go from the centre. Put e for the radius of W, £ for 

 the number of cogs on the edge of the disk X ; and rj for 

 the number of cogs on the spindle B ; we have 



Speed of X : speed of B =77 : £, 

 » W: „ X=8:e, 

 „ W: „ B=8,:ef=l:*, 



which gives us k= ~. 



If we place the arm G in any other position and turn the 

 spindle B for some time, and then read p and q as the increase 

 in the readings of the counters of the adjacent cardinal points, 

 the resultant reading for magnitude is \/p 2 + q 2 , and the 

 number of revolutions of B will be k\/p 2 + q 2 , w T here k has the 

 same value as before. The direction of the bar Gr may be 

 determined by taking 6 as the angle between the bar and the 

 cardinal point to which p relates. Then 



p : (7 = cos 6 : sin 0; 

 whence 



tan0=£. 

 9 



Of course if B is rotated a certain number of times (say 500), 

 the value of \/p 2 +. q 2 . must be the same in whatever position 

 G has been fixed throughout the experiment. 



By repeating this experiment with different positions of G, 

 we get a means of testing the amount of error to which the 

 instrument is liable. 



When the anemometer is connected with a vane and with 

 Robinson's cups, let us suppose the wind to remain in one 

 direction for some time; or, if that cannot be secured, let us 

 either tie up the vane, or disconnect it from the spindle A 

 and fasten the spindle. Let I revolutions of the cups indicate 

 one mile of wind, and m revolutions of the cups give one revo- 

 lution of the spindle B. Let p and q be the readings of the 

 counters, then the miles of wind indicated will be 



I 



ks/p 2 + q 2 .— ■ 



In ordinary use the bar G will be shifted about in all direc- 

 tions by the wind, and we may have an increase in the read- 

 ings of all the counters; let these be represented by n } s, e, 



