TORSIONAL OSCILLATIONS OF WIRES. 435 



Two of these curves are shown in fig. 16. The part of a curve which corresponds to 

 the outward motion can easily be distinguished from that which corresponds to the 

 inward motion by its greater amplitude. In the first curve, 20 semi-vibrations take 

 place in the range AB in the outward motion, while 20 take place in the range CA in 

 the inward motion. The difference BC corresponds (allowing for the slight difference 

 at the end A) to about one-third of a semi- vibration. Thus the outward motion over the 

 range AC occupies less time than the inivard motion over the same range, the difference 

 being about 1 in 60. 



Result of Heating the Wire to Redness. 



[Added 18th July 1898. — It is to be expected that the molecular freedom which is 

 introduced by heating the wire to redness will undo, to a great extent at least, the 

 effect of fatigue. Before testing this point the wire was subjected to greater fatigue 

 than on any previous occasion, and an experiment was then made on the date 1.7.98. 

 The results were 



rt = 4, « = 1-015, & = 89-6, nb = 91, y -36:7-. 



Thus by excessive fatigue the value of b was made smaller than it had ever been, while 

 n, as formerly under such conditions, approximated to unity. 



On the date 14.7.98 the wire was heated to redness by a Bunsen flame, the lead 

 ring being removed to prevent stretching. An experiment was then made, and the 

 results were 



« = 7, 72 = 1-253, 6 = 680, nb = 852, z/ = 434. 



A comparison with the results given in the last column of Table IV. shows that b has 

 become much more than twice as large as the greatest previous value. 



It is interesting to compare this result with the results of two experiments made on 

 the date 19.7.93, but not published in the first paper. In these experiments the wire 

 hung inside a long solenoid composed of two similar coils of stout copper wire. In the 

 first experiment a heavy current was run, in opposite directions, through the coils. The 

 effect was to maintain the wire at a temperature of about 80° C. The results were 



ffl = 2, h= 1-747, A = 536, w5 = 936-. 



The difference between the conditions now considered and those above described is that, 

 now the wire is maintained at a comparatively high temperature during the experiment, 

 while formerly it was heated to redness and was then experimented upon when cold. 

 Though 6 is not quite so large in the latter case as in the former, n is considerably 

 greater than formerly — so much so that nb is greater in the case now under discussion 

 than in the other. Hence, when the temperature is maintained high, the loss of energy 



