266 ELECTRICAL MEASUREMENTS 



no longer applicable. In this case, experimentally determined 

 values, / (see Fig. 150), must be used. They are deduced on 

 the assumption that at spark-over, g 8 is as given by Fig. 151. 



Variations of frequency, at least up to 1,000 cycles, have no 

 influence on the results. 



As an example of the use of the foregoing, the voltage necessary 

 to break down a 10-cm. gap between spheres 25 cm. in diameter, 

 neither sphere being grounded, will be calculated. Assuming 

 the barometric pressure to be 760 mm. 



/ 0.54 \ w 



0. = 27.2(1 +-7= - =31.4 



cm. 



*- 10 -08 

 a " 12.5 ' 



/ = 1.283 by table. 

 31.4 10 

 ~~ = max ' 



For a sinusoidal wave V = 173 kv. effective. 



If the spheres are 25 cm. apart and one of them is grounded, the 



calculation is 



? - 25 - - 2 

 a 12.5 " 



/o = 2, from Fig. 150. 



01 A . o^ 



/. V = - - = 392 kv. max. 



A 



For a sinusoidal wave V = 277 kv. effective. 



Reference to Fig. 151 shows that in important work it is best 

 to use large spheres and thus avoid the use of the steep part of 

 the curve and the consequent uncertainty in g s . 



In practical work it is best to take the spark-over voltages from 

 experimentally determined curves, or the table given on page 

 267, rather than to use the above formulae. 



The use of a spark gap is not without danger to the 

 apparatus under test; for high voltage surges may be set up 

 when the gap breaks down, hence the use of the current-limiting 

 resistances. 



