1921] on Elasticity 385 



clamped to a fixed support carried on a stout glass rod. The upper 

 end was clamped to a stiff hardwood rod, wood and glass being 

 chosen as being bad conductors of heat. 



Mounted in this way the rod could oscillate (like one end of a 

 tuning-fork), and being itself so stiff compared to the test-piece as 

 to remain practically unbent, the force of restitution was determined 

 by the elasticity and dimensions of the test-pieces only. The elastic 

 quality called into play when a bar or plate is bent is Young's 

 Modulus, and hence to determine the variations of Young's Modulus 

 at different temperatures it was only necessary to compare the periods 

 of oscillation of the test-piece at those temperatures, the temperature 

 being the only condition which was altered (if the small correction 

 due to dilatation or contraction is neglected). The actual dimen- 

 sions of the test-piece and the periods do not matter, all that is 

 required being the comparison of its periods at different tempera- 

 tures. In order to make this comparison the actual periods had to 

 be measured with considerable accuracy. 



The periods employed were generally between half a second and 

 two seconds, and if the rod is displaced and then left to itself the 

 oscillations so started would from various causes die out before a 

 sufficient number could be counted to give an accurate result — 

 accurate, that is, to one part in two or three thousand. The oscilla- 

 tions therefore were maintained by an arrangement I had before 

 used for an electric clock. This gave an impulse to the rod at the 

 moment of its passing through the zero position. An impulse given 

 to an oscillation at that time has no effect on the period, but merely 

 increases its amplitude if there is no resistance, or if there is resist- 

 ance keeps the amplitude constant by adding at each impulse the 

 loss sustained during the previous beat. 



It is shown in the slide (7). (Description.) 



I have the maintainer here, but as its action could not be well 

 shown when applied to experiments described, I have arranged it to 

 maintain the vibrations of a pendulum the image of which will be 

 thrown on the screen. 



The current which worked the magnets of the maintaiuer was 

 also used to record each oscillation on a chronograph on which a 

 clock recorded seconds of time. In this way time could be measured 

 to about gVth of a second, and since from 100 to 300 or more oscil- 

 lations were generally made by the rod in each experiment, the 

 period could be well determined. 



It was observed that the more fusible was the metal subjected to 

 trial the greater was the variation of the elasticity with change of 

 temperature, the change being always an increase of the value of 

 Young's Modulus with a decrease of temperature. 



The metals tested were (in the order of their melting point), 

 rhodium, platinum, iron, palladium, nickel, copper, gold, silver, 

 magnesium, aluminium, zinc, lead, cadmium, bismuth, and tin. 



