344 O. JBarus — Resistance of Stressed Glass. 



Hence in an extremely unfavorable case, the resistance effect 

 due to elastic change of dimension (—-30 per cent) is only 

 about -§- of the observed effect of traction (— 1/4 per cent) pro- 

 duced by a pull much below the tenacity of glass, the said 

 pull (18 kg.) being certainly not more than \ the maximum 

 load. Hence these effects are very different, and it follows 

 that the decrement actually observed is principally due to 

 decreased molecular stability superinduced by stress. In ecpia- 

 tion (1), s is therefore the variable which chiefly responds to 

 the action of stress. 



To obviate the troublesome occurrence of 8R" j R, the 

 column of mercury in most of my experiments was made so 

 long as to extend far above the zone of conduction of the 

 stretched glass tube (see figure 1). In the apparatus for steam, 

 figure 2, the menisci of the column are advantageously raised 

 quite above the cork. In such a case 8R /, /R=0, and the 

 elastic discrepancy is simply —81/1. 



In one respect this reasoning is deficient. It does not take 

 into account the changes of elastic behavior of glass clue to the 

 heating to 190°. Tabulated constants for this large interval 

 are not available.* Hence special cathetometric measurements 

 must be made. At 190° this is difficult, and for these and the 

 other reasons given above, §5, it is expedient to refer to the 

 complete set of measurements at 100°. § 6. 



6. At 100° the results can be made more accurate than the 

 above chiefly for two reasons. In the first place the tempera- 

 ture is easily obtained absolutely constant ; in the second elas- 

 tic changes of dimensions can be directly measured with facility. 

 In Table I, I have given the results obtained with the appara- 

 tus, figure 2. The method of measuring these large resistances 

 (glass at 100°) is necessarily chosen more delicate than above. 

 I used a high resistance Thomson's galvanometer read off by 

 Hallock's short range telescope, and adjusted for differential 

 work. The needle being practically ballistic in kind, the 

 maximum deflections (swing) obtainable by alternately adding 

 and removing the loads R, were used for comparison (method 

 of multiplication). I then determined the amount of oscilla- 

 tion produced by inserting known resistances into one or the 

 other coil of the differential galvanometer. Knowing the 

 resistance of each tube (mean values) from special and pre- 

 liminary measurements, I was able to deduce the percentage 

 variation of the resistances of glass across the lines of stress. 

 Table I contains four series of these experiments, i. e. two 

 sets of results for each pair of tubes. R, the observed electri- 

 cal resistance per tube was found to be about 7,500,000 ohms 



* Kohlrausch and P. E. Loomis (this Journ., II, 1, p. 350, 1870), give low tem- 

 perature data for metals. 



