IN TORSIONAL OSCILLATION 121 



thus the accuracy could not be so great as in the former 

 experiments. This was found to be even more apparent in 

 the next experiment, with a wire of diameter 2*0 mms. It 

 was found impossible to take readings with any degree of 

 accuracy with 12 inches of this wire, and so a double length 

 was taken. Distinct pendulum oscillations were also of 

 more frequent occurrence when the thick wire was used, and 

 the curve obtained showed a wavy appearance. By drawing 

 a straight line through the observed points, so as to eliminate 

 the disturbing effect, the following results were obtained : 



a=90, w=-84, 6=427. 



This value of 6, as shown on page 117, can be compared 

 with the values got with the wires already discussed. For 

 it was shown that in a wire of half length 



Thus the value of b will be got for that of double length by 



6=6'-2" 

 =238. 



The value of 6, then, although not exactly under control, 

 is seen to decline steadily with increases of diameter, thus : 



The values of a and n, then, are largely independent of 

 change of sectional area, just as they were of change of length. 

 This is to be expected, in accordance with the original assump- 

 tion that the loss of energy for a given material depends solely 

 on a power of the angle of distortion. 



Q 



