15° 



15° 



20° 



3-7 



3-7 



•40 



4-6 



4-8 



6-5 



92 C. Barus — Tsothermals, Isopiestics and 



can only infer that its value is of the order of -005, and that 

 it bears no obvious relation to the initial viscosity, or to tem- 

 perature. 



In endeavoring to improve upon this work I cut a screw 

 thread in the inside of the transpiration tube and thus largely 

 obviated slipping, by compelling the charge to flow on itself. 

 I also made all observations in triplets, including each measure- 

 ment at a high pressure or temperature between two fiducial 

 measurements at a given lower pressure or temperature. Only 

 such observations were taken for which the fiducial data were 

 identical. Finally by treating the charges individually, I 

 found that although the viscosities at the same temperature 

 were very different, the pressure coefficients followed each 

 other in the order of the initial viscosities. 



Experiments made in this way showed — 



Temperature = 10-5° 13-9° 11-9° 180° 

 ?7 /l0 9 = -50 2-] 5 2-50 "70 



&X10 3 = 4-2 4-3 8-9 5-3 



These are the best results I have been able to obtain. How 

 nearly linear the variation of viscosity with pressure is, may 

 be seen in the following example of consecutive measure- 

 ments : 



Pressure excess Ap = 340 700 1055 1410 1770 340 atm. 

 Viscosity, ?//] 9 ' = -87 1'20 1'60 2-07 2-30 -87 cj/cs 



Even in case of high viscosity (>10'°), a tendency of vis- 

 cosity to increase at an accelerated rate with pressure is only 

 vaguely apparent, e. g., 



Pressure excess, Ap = 505 1020 5U5 1540 1022 1540 atm. 

 Viscosity, 7//10 9 = 8-4 12-6 7-9 17*9 12-9 17*1 gj.cs 



Taking the above work as a whole, therefore, I am bound 

 to infer that within the range of observation (2000 atm.), the 

 pressure coefficient is constant: for though varying between 

 •004 and "009, it shows no discernable relation to the initial 

 viscosity (^ for Ap ~ 0), or to temperature. In other words, 

 to assume that the rate at which viscosity increases with pres- 

 sure at any temperature, is proportional to the initial viscosity 

 at that temperature, is the nearest approach to the actual state 

 of the case which my observations enable me to make. Tak- 

 ing the mean of all values in hand I thus obtain — 



where the subscripts show the temperature (0) and pressure (p) 

 at which viscosity {rj) is taken ; and where p is the mean of the 

 pressures at the two ends of the transpiration tube. 



