its Solvent Action on Glass. 115 



Some allowance must however be made for the attenuated 

 thread of mercury (§ 3). If v/Vbe plotted as a function of 

 p, a series of curves is obtained as shown in figure 1. Consid- 

 ering the difficulties of measurement, they are satisfactorily 

 regular, Temperature and time are affixed to each curve. In 

 table 3, (v/ V) is rigorously the ratio of increment of length 

 to the original length at z.4°, due to thermal expansion and 

 concomitant chemical action. The radius of the tubes widens 

 as solution proceeds ; but the datum (v/ V) suffices for the 

 present purposes. Let /? be represented in its dependence on 

 (v/ V). The plotted curve is a line of remarkable regularity, 

 as shown in figure 2. It follows from the chart that /? in- 

 creases 11/10 6 for each per cent of volume decrease of the 

 water undergoing silicification. This is about 75/10 6 per centi- 

 meter of length, agreeing substantially with the former result. 

 Again v/ V decreases L3 per cent for the interval of observa- 

 tion of 42 m , or about *3 per cent per minute, thus again agree- 

 ing with § 6. See figure 3. 



Suppose the line for /? and (v/ V) to be prolonged as far as 

 ( v / V) = 130/10 3 , which holds for time =0. The datum for fi 

 so obtained, ought to give me the normal compressibility of 

 pure water at 185°. Making the prolongation, however, I 

 find an excessively small result /3=50/10 6 nearly. This merely 

 shows, since chemical action is very rapid, that the time at which 

 it commenced is only roughly indicated. It is probable never- 

 theless that /? 186 will not be greater than TO/10 6 . Hence even 

 above 100° the compressibility of water increases at a very low 

 rate with temperature ; at a rate about J that of paraffine, 

 for instance. Cf . figure 4. I think this indicates exceptional 

 stability of the water molecule. 



9. Now what is the underlying cause of the action described ? 

 Clearly I think, an instability of the glass molecule at 185°, 

 much rather than any instability of the water molecule. This 

 is an accordance with the evidence I adduced in studying the 

 electrolytic conduction of stressed glass,* and corresponds also 

 to the diminished viscosity of glassf at the stated temperatures. 

 At 185° the cohesive affinities^: of the water are sufficient to 

 disintegrate the glass molecule. 



The increase of /? with time must be due to the solution of 

 silicate. Indeed it would be difficult to devise an experiment, 

 in which the progress of the continued solution can be so well 

 discerned as is possible in the present incidental results. I am 



* Barus: This Journal, xxxvii, p. 339, 1889. 



f Barus and Strouhal : This Journal, xxxi, p. 439. 18S6; ibid., xxxii, p. 181, 

 1886. 



\ A term which will be defined succinctly in the course of the present series of 

 papers. 



Am. Joint. Sol— Third Series, Vol. XLI, No. 242.— February, 1891. 

 8 



