Glass, and Metal Wires subjected to Constant Pull. 521 



carrier \Y which had a system of vanes V dipping into oil to 

 damp vibrations. 



Fi£. 1. 



o 



f 



R 



P^H 



To the upper end was also attached a circular 

 piece of brass C from which depended three 

 rods, a. of glass, and to the lower ends of a a 

 ring of brass R was fixed. An optical lever 

 rested with two of its legs on the ring R and 

 one on the table T. A telescope viewed the m — r ^Jh 

 reflexion of a millimetre-scale in the mirror of 

 the optical lever 0. 



This arrangement is that used by Mr. G. 

 A. Shakespeare in this laboratory, the rods a 

 serving to compensate for temperature altera- 

 tions and for any give of the support. The 

 length of the optical lever was '820 cm. and 

 the distance of the scale was 3 metres during 

 some experiments, and 461 cms. in others. 



In the same way as for the indiarubber ex- 

 periments the weight was put on by hand as 

 gently as possible, and the reading in the tele- 

 scope taken at intervals. At first no consistent 

 results could be obtained, but when the fibres 

 were put on one side for a few weeks after their t\* 



manufacture the subsequent results for stretch- 

 ing were quite concordant. 



In the case of glass, as in that of indiarubber, 

 the law of stretching was 



as = a + b log t , 



the symbols having the same significance as 



before ; and when the load was left on for a 



long time and then removed the creep back was 



proportional to the logarithm of the time which 



had elapsed since the removal of the load if 



the duration of the observations was short compared with 



the time since the load was put on. 



The numbers given (p. 522) were obtained for a load of 

 779 kilograms per sq. cm. cross section. 



Here again, for the first few numbers x — a is rather small, 

 and so a small error in x would make a large error in x — a. 



a was found as before by plotting the stretch against log t. 

 and producing the straight line so obtained till it cut 

 the axis, log t = 0. The value of x at log £ = is of 

 course equal to a. If x is measured in scale-divisions and 

 / in minutes, we see that b in this case is equal to 2*74 : if x 



Phil. Mag. 8. 6. Vol. 9. Xo. 52. April 1905. 2 M 



