296 A. W. Witkowski on the ■ 



employed. The usual oiled leather piston is replaced by 

 one packed with fibrous asbestos, and lubricated by a very 

 small quantity of water (a device used also by Wroblewski). 

 Although this arrangement renders the working of the pump 

 exceedingly tedious, nevertheless I found it absolutely neces- 

 sary, as the oiled piston yielded a very impure gas. 



The air was taken from the outside of the laboratory ; it 

 passed through several washing-bottles filled with solutions of 

 potassium hydroxide and sulphuric acid. After condensation, 

 it was dried in the manner described above. The air treated 

 in this way was submitted repeatedly to chemical tests, and 

 proved to be sufficiently pure and dry. 



§ 6. Determination of & 1 and s 2 . — Before commencing the 

 experiments, it was necessary to determine very exactly the 

 capacities of the bulbs s 1 and s 2 . This was done by mercury- 

 weighings at the temperatures 0° and +100°. The true 

 capacity during experiment depends on the temperature of 

 the bulb and on the pressure employed. 



To calculate the capacity at low temperatures, I used the 

 coefficient of expansion of glass, determined in the interval 

 0° to +100°, and corrected it in accordance with the expe- 

 riments of I. Zakrzewski on the expansion of solids at low 

 temperatures *. 



The influence of pressure on the capacity of the bulbs has 

 been determined directly by submitting them to known pres- 

 sures while filled with well-boiled mercury. From the observed 

 variations of position of the meniscus in the calibrated stem, 

 it is easy to calculate the variations of capacity with the aid 

 of Tait's and of Amagat's determinations of the compressibility 

 of mercury. In the limits of pressure used this dependence 

 was a linear one. Denoting the corresponding coefficient by 

 x, we may write 



s p =s(l + xp). 



For a pair of bulbs of very unequal capacity (1 : 2) , I have 

 found ^=0-0000062 and ^=0*0000064 (per atmosphere). 

 These values refer to ordinary temperatures. As there is 

 nothing known about the variations of elasticity of glass at 

 very low temperatures, I have always used the above value 

 of a?; I consider the error resulting from this proceeding as 

 altogether insignificant, the more so as the correction itself, 

 depending on pressure, is very small. 



§ 7. Mode of Experimenting. — I will now describe the 

 mode of experimenting and calculation of the results. 



* Transac. Cracow Acad. vol. xx. (math, class.). 



