528 Mr. Phillips o?i the Slow Stretch in Indiarubber, 



The curve connecting b with the load for silver is given on 

 Curve No. III. d. All these curves show the characteristics 

 mentioned in the case of copper. No great importance mast 

 be attached to the absolute value of these numbers, for besides 

 the great effect which is produced by the annealing tempera- 

 ture, by the length of time during which the annealing- 

 continues, and by the rate at which the wire is cooled down, 

 there is an effect produced by the way in which the weight 

 is put on. Added to this, different specimens of wire give 

 very different results, though when pure wires were obtained 

 from Messrs. Johnson & Matthey these differences were fairly 

 small. 



A fundamental difference between the slow stretching of 

 indiarubber and glass and that of metal wires is that the former 

 is reversible, i. e., the creep back when the load is removed 

 follows the same law, while in the case of the metals there is 

 no appreciable slow recovery. 



In the case of the stretching of these metals, it is just as 

 difficult to make a reasonable assumption which shall give the 

 law as it is with indiarubber. If we assume that the perma- 

 nent stretch is produced by cleavage of crystals, we require 

 that the rate of breakdown of the crystals shall be given by 



dx 

 dt 



where x is the stretch. Or, as it seems reasonable to suppose 

 that the stretch is proportional to the number (n) of crystals 

 broken down, 



dt 

 or, if N be the number of unbroken crystals, 

 dn 

 di 



All previous assumptions have led to a limit, or else have 

 attributed the characteristics of a liquid, and so have made 

 the stretch a linear function of the time. 

 We see, however, that 



d?x Rr dx 



W = ~ a ^dt 



- z=ae~ Bx 



Y is proportional to I ^r ) . 

 We might therefore. assume that after the very beginning of 



i.e. -— 



dt 



