C. Barns — Resistance of Stressed Glass. 341 



or with a lever arrangement for twisting. In the experiments 

 made the load was gradually increased as far as 20 kg., but the 

 tubes were strong enough (theoretically) to sustain about three 

 times this weight. The remainder of the figure shows the 

 devices for heating and for passing the currents. Figure 1 is 

 adapted for high boiling points (aniline, etc.), figure 2 for 

 steam. An apparatus similar to figure I was used for mer- 

 cury. In figure 1, G is the ebullition liquid, heated by a Gibb's 

 ring burner RR surrounding the wide glass tube old. A 

 narrower glass tube gg, closed below with a perforated cork 

 through which pass the experimental tubes abc and ekh, is 

 partially filled with sodium amalgam. This is practically one 

 terminal of the battery, the wire connecting at p. The other 

 terminal, after passing through the respective coils of a differ- 

 ential galvanometer, connects at m and f with the sodium 

 amalgam contained in the experimental tubes abo and ekh. 



The notation in figure 2 is the same as that in figure 1. The 

 two forms of apparatus are essentially identical, except that in 

 figure 2 it is expedient to pass steam through dd, the vapor 

 entering at S and leaving the apparatus at S'. For reasons 

 stated below, §5, it is desirable that the menisci of the 

 amalgam contained in ahcfrand ekhf, figure 1, be visible above 

 the level of the upper cork of the tube dd. The amalgam in 

 gg, figure 1, should be submerged below the level of G. Inas- 

 much as sodium amalgam is only necessary at the anode, 

 ordinary mercury may be used at cathodal parts ; and these 

 may therefore be exposed to hot air or steam without annoy- 

 ances. To summarize : current arriving at m and f passes 

 into the sodium amalgam core of the tubes, abc and ekh, thence 

 across the walls of the hot parts of these tubes into the mer- 

 cury surrounding them, and finally via jp back to the battery. 

 Regarding other apparatus, cf . § 7. 



3. I commenced work with torsion exjDeriments of which I 

 may indicate something here. A battery of 10 Grove cells was 

 used and the aniline at G, figure 1, kept both below and at the 

 temperature of ebullition. The deflection of each coil alone 

 being 16"4 0m , it was found that the differential action (nearly 

 zero) could not be modified by twisting more than •02 cm . 

 Hence the specific electrical effect of twisting can not be 

 greater than about '1 per cent. The resistances encountered 

 in these cases were about 200,000 ohms. Profiting by this 

 preliminary experience however I was ultimately able to de- 

 tect and measure the effect of torsion on electrolytic conduc- 

 tivity, using a different and more sensitive method to be indi- 

 cated below, §7. 



4. In case of traction, the data decisively indicated an 

 increase of conductivity proportional to the pull. But this 



