342 C. Barus — Resistance of Stressed Glass. 



result is necessarily complex in kind, and must be carefully 

 scrutinized before its true signification can be stated. I will 

 therefore give my experiments in chronological order, the first 

 series being made at 185° (aniline), the second at 100° (steam), 

 the third series finally at 360° (mercury). 



One remark may be made at the outset: inasmuch as the 

 electrical effect of traction is persistant with the traction, and 

 is an increment of conductivity, it can not be due to tempera- 

 ture. For the extension of an elastic solid like glass* produces 

 temporary cooling, § 8. 



5. The resistance across the walls of the experimental tube 

 at 190° was about 100,000 ohms. In case of intense ebullition, 

 the temperature is not fully constant. It is therefore desirable 

 to use the apparatus, figure 1, just below the boiling point of 

 aniline, and to bring the plane of the ring burner slightly 

 below the plane of the ebullition liquid G. Parts of the ap- 

 paratus which are not to be heated are screened with asbestos. 

 In this way a nearly stationary distribution of temperature is 

 reached. 



Under these circumstances, when a weight of 18 kg. is alter- 

 nately placed on the scale pan and removed from it by 

 mechanism, thus subjecting the tube to periodic pulls of the 

 force given, a definite and persistent oscillation of the galvano- 

 meter needle ensues synchronously with the period of stress. 

 The amount of this oscillation was found to be equivalent to a 

 resistance-decrement of 1500 ohms for the stressed tube. In 

 other words the effect of the pull of 18 kg. is a diminution of 

 the resistance of the stressed tube amounting to about 14 per 

 cent. These experiments were repeated many times with prac- 

 tically the same results, e. g. 



P = 2 kg., resistance reduced .4 per cent. 

 P = 15 kg., " " 1-2 " 



P=20kg., " " 1-4 " 



data in which the oscillation of the needle was made the basis 

 of comparison. They betray a somewhat wide margin of 

 error, because glass at 190° is exceedingly sensitive even to 

 trifling changes of temperature. Nevertheless the data are 

 sufficient for the present purposes ; and work of a more precise 

 character with high temperature vapor baths seemed to me to 

 be superfluous. By using a more sensitive galvanometer such 

 measurements can be repeated at 100° with facility and much 

 greater precision. 



6. The result obtained is clearly a superimposed effect, being 

 due in part to the elastic change of dimensions during stretch- 

 ing, and in part to the direct action of stress in promoting 



* Cf. Sir William Thomson's collected papers, vol. i, p. 308. 



