﻿Air Velocity by means of Hot Wire Anemometer. 561 



It is preferable to calibrate the wire directly by means of 

 a rotating arm, the velocity being corrected for "swirl " in 

 the manner already described *. Wires are usually calibrated 

 by the writer over the range V = 60 to 800 cm./sec, 

 enabling the constants of the formula 



? = t 2 + V V (1) 



to be determined ; experiments are discussed below showing 

 that this formula may with fair accuracy be employed in 

 the determination of velocity very considerably above and 

 below these limits. For precision work a series of about 

 ten determinations of current and velocity should be taken, 

 and the points y — i 2 and x— y/V plotted on accurate section 

 paper in order to eliminate possible gross errors or accidental 

 .mistakes of reading. The line of closest fit to this series of 

 points should then be determined by calculation ; the scale 

 is altered by multiplying y or x by a suitable power of ten, 

 so that these coordinates are expressed by numbers of the 

 same magnitude. Under these conditions the line of closest 

 lit may conveniently be taken to be the major axis of inertia 

 of the' system of points regarded as masses of equal weights. 

 If there are n points, the inclination of this line, whose 

 equation may be written y=y + x tan (9, is easily seen to be 

 given by the formula 



^(x~-y 2 )/n-(x 2 -y 2 ) 



,i: and y being the coordinates of the centre of gravity given 

 by x = ^%{x)jn, y = 2,(y)/n. From this formula we find tan 

 and y , from the formula y^—y — x tan 6 ; hence reducing 

 back to the original scale we obtain i 2 and k. The calcula- 

 tion is somewhat facilitated by taking n = 10. The agreement 

 of two independent determinations of the calibration constants 

 computed in this way is well illustrated in the case of 

 wire 17, Table II. 



While velocities corresponding to an observed value of i 

 may easily be obtained from the calibration curve correspond- 

 ing to formula (1), a considerable saving of time can be 

 effected if more than 100 current observations have to be 

 reduced to velocities by calculating out a conversion-table 

 corresponding to the formula 



V=*-V-*o s )', (3] 



.at intervals of 0*0 L ampere over the range required. The 



* Phil. Trans, paper, pp. 388, 428 ; in the sequel this paper will be 



referred to as reference (1 ). 



Phil Mag. S. 6. Vol. 29. No. 172. April 1915. i ; 



