its Solvent Action on Glass. Ill 



Amagat* applying a new method of pressure measurement " a 

 piston libres," operates with hydrostatic pressures as remarkably 

 high as 3000 atm., and at temperatures between 0° and 50°. 

 He shows among other results relating as yet chieiiy to thermal 

 expansion, that the compressional peculiarities of the behavior 

 of water vanish at high pressures and increasing temperatures 

 (interval 0° to 50°), thus further corroborating Grassi. Many 

 data are given ; but the research is unfinished. Taitf in a 

 final paper summarizes much of his work and begins a series 

 of experiments showing that the effect of solution is analogous 

 to an increase of internal pressure. A critical revision of 

 earlier work on compressibility may be found in Tait's " Prop- 

 erties of Matter." 



From this brief summary it appears that results anticipating 

 the contents of the present paper are not in hand. There is 

 another class of experiments relating to the expansion of water 

 in glass tubes to which I must advert. WaterstonJ published 

 a very full series of results carrying the work as far as 300°. 

 He was annoyed by the action of water on glass, but does not 

 further consider it. For very high temperatures the experi- 

 ments of Daubree§ and others, are well known. 



3. In the present work pressures were applied by aid of 

 Cailletet's large force pump. The thread of water is enclosed 

 in a capillary tube, between two end threads of mercury, 

 and the distance apart of the two inner menisci, corres- 

 ponding to any given temperature and pressure, measured by 

 Grunow's cathetometer. The tube, suitably closed above, is 

 exposed in a vapor bath (boiling tube). At 185° (aniline), the 

 thread of water soon loses its transparency, becoming white 

 and cloudy. Fortunately the siliceous water is translucent. 

 By placing a very bright screen behind it, the demarcation 

 between water and mercury remains sufficiently sharp for 

 measurement. After the action has continued for some time, 

 say an hour, the columm is solid at high pressures (300 atm.), 

 though it is probably only partially so at 20 atm. In conse- 

 quence of this, threads of mercury break off during advance 

 and retrogression of the column. Further measurement is 

 therefore not feasible. Toward the close of the experiment, 

 moreover, the mercury thread is pushed forward, enclosed by 

 walls of semi-solid siliceous water. The thread is therefore 

 of smaller diameter and the measurement correspondingly 

 inaccurate. 



In obtaining these data, I followed a customary routine of 

 increasing pressure from zero to the maximum, then decreasing 



* Amagat: C. E., ciii, p. 429, 1886; ibid., civ. p. 1159, 1887; ibid., cv, p. 1120, 

 1881. 



•j-Tait: Challenger Reports, ii, part 4, 1888. 



% Waterston: Phil. Mag., (4) xxvi, p. 116, 1863. 



gDaubree: Et. synthet. d. Geol. exper., 1879, p. 154 et seq., Pari?, Dunod. 



