C. Barus — Resistance of Stressed Glass. 347 



made cathetometrically, and are not intended to give more 

 than a safe estimate of the elastic effect in question. The 

 glass tube to be examined was surrounded conaxially by a 

 second wide tube of glass, through which steam at 100° con- 

 tinually circulated. Measurements were also made at 16°. L 

 is the length between fiducial marks. 



Utilizing these values to obtain a superior limit of the elastic 

 discrepancy in Table I, it appears that dl / 1<^ ' 0004 and 

 Sp/p<i • 0001, these data being the largest obtained for the largest 

 load, 18 kg. Hence by equation (2), -dR'/R<2-5 x -0004= -0010. 

 In other words the elastic discrepancy is numerically much less 

 than "1 per cent of R, whereas the corresponding mean value 

 for the traction effect in Table I (apparatus with tubes I and 

 II, low menisci) is *75 per cent. Again for raised menisci 

 (Tables I and II, tubes III and W)' i -d/R , /R=dl/l='0004t. In 

 this case the corresponding mean value of the traction effect is 

 numerically greater than "50 per cent. In both instances it may 

 be safely inferred that error introduced by elastic change of 

 dimensions is at most about 1/10 of the decrement of resist- 

 ance actually observed as the effect of stretching. 



9. I will make a final consideration here, relative to temper- 

 ature. The thermal effect of traction is negative, its influence 

 on R must therefore be a resistance increment, i. e. opposite 

 in sign to the effect observed. Nevertheless, it is desirable to 

 obtain some estimate of its value, which will probably be 

 found too small for direct measurement. Since P=20 kg. and 

 SLjL<^ • 0004 cm , the total energy elastically potentialized per 

 linear centimeter during stretching is RdL/Z<^10000 ergs. 

 Hence, even if all this energy were converted into heat, the 

 increase of temperature resulting in case of the given tubes 

 (section -10 om2 , density <^3, sp. heat </2) would be about 

 107240x1 O 4 ; i. e. less than -005°. This datum is too small to 

 produce serious error even in consideration of the phenomenal 

 sensitiveness of hot glass to temperature variations. Estimat- 

 ing that the resistance of glass decreases several per cent (5 to 

 20) per degree, between 100° and 200°, the thermal discrep- 

 ancy can not be greater, numerically, than the elastic dis- 

 crepancy. 



10. I have now to communicate the data obtained at 360°. 

 This case possesses some points of special interest, because the 

 differential apparatus is itself a battery, the action of which 

 enters in a complex manner. The electrolytes here are the hot 

 glass tubes containing amalgam and surrounded by mercury. 

 The actual apparatus was a simplified form of figure 1. Figure 

 3 presents a clearer diagram of parts, in which a and b are the 

 hot glass tubes in question, E the battery and D the differen- 

 tial galvanometer. The electrical currents due to E are indi- 



