Chemical Equilibrium of Solids. 15 



safety of these methods and replaced them by simple circuit 

 methods. The errors to be guarded against are primarily in 

 the experiment itself. Again, the resistance-tube (figs. 1-3) 

 is well adapted for simple circuit-work, since the electro- 

 motive force NaHg/glass/Hg may exceed a volt, and the 

 only pronounced resistance is in the cell. Only at 100° need 

 auxiliary cells be inserted. 



Throughout the work the low-pressure or fiducial resistance 

 is measured before and after the high-pressure resistance. 

 This safeguard is necessary. From three to Rye minutes were 

 allowed for the elimination of thermal discrepancies. The 

 fiducial resistance is to be measured under a pressure of 100 

 to 200 atmospheres, both to condense the air retained in the 

 pump and to make the oil less compressible (§ 32) . Not much 

 care was taken in measuring the absolute resistances, since 

 the inferences of this paper are to be drawn from the in- 

 crements. 



8. In the high-resistance experiments I used the Elliott 

 square pattern of Thomson's high-resistance astatic galvano- 

 meter. The readings, however, were made by PoggendorfFs 

 method of telescope and scale. To adapt Thomson's concave- 

 mirror adjustment for this purpose, I compensated the cur- 

 vature by an ordinary " No. 36 " concave spectacle-glass fixed 

 in front of the mirror, diffuse light being suitably screened off. 

 Glass " No. 36 " slightly under-compensates the mirror, 

 leaving some magnification which is not undesirable. In 

 how far a weak telescope may be made to do the service of a 

 more powerful instrument, in virtue of the concave mirror and 

 a lens even weaker than No. 36, remains to be seen. With 

 the above adjustment of lens and mirror, using a telescope 

 magnifying somewhat over 20 diameters, and a scale some- 

 what further than 2 metres from the mirror, I obtained an 

 image about 4*5 centim. long and sufficiently distinct for all 

 practical purposes. Thus, if one volt act in one million 

 megohms, the current produced would only just escape 

 detection. 



At 200° the galvanometer was suitably shunted. 



9. Regarding other manipulations, the paper cited in § 2 

 will have to be consulted. Electromotive forces were mea- 

 sured by a zero method. 



Observations. 



10. Table I. contains data for the apparent change of the 

 resistance of glass with pressure, when the liquid which 

 transmits pressure is ordinary saponifiable sperm-oil. Here 

 6 denotes the temperature of the vapour -bath, approximately, 



