499 



the melting-point pressure of tiie helium-thermometer with meicury- 

 maiiometer was 25.738 cms. mercury, that of the helium-thermometer 

 with heated wire manometer 5.240 ems. In the second series these 

 pressures were 25.358 cms. and J. 2059 cm. mercury respectively. 

 The vapour-pressure measurements were conducted in the same 

 manner as before (comm. 119 and 1246j. The results were as follows 

 (uncorrected T stands for T not corrected for B and for thermal 

 molecular pressure'): 



The first column gives the vapour-pressure of helium at the 

 corresponding temperature, the second the temperature as calculated 

 with the aid of the ordinary gas-laws (with B = 0). The great 

 difference between the temperatures found in this way with the two 

 thermometers is very striking, especially in the last series of meas- 

 uremeuts in which the melting-point pressure of the second thermo- 

 meter was very low. The influence of the thermal molecular pressure 

 causes a temperature of 5°. 5 to be found instead of 4°. 2. 



5. Correction for tlir tlunnnal molecular pressure. Expressions for 



the thermal molecular i)ressure which are valid for the ranges 



2R 2R 



0< — <1 or 10 < — < CO have been developed by Knudsen. It is 



clear, that the choice of the two limits 1 and 10 has been somewhat 



arbitrary, but we may assume, that, when these limits are attended 



to, the uncertainty of the results of calculation by means of these 



formulae, supposing the constants which occur in them to be known, 



is on the average smaller than IV»- Knudsen's formulae do not hold 



2.R ^ 

 for the intermediate range of 1 < — < 10. 

 "" X - 



The condition of pressure-equilibrium in a tube with a gradient 



of temperature is in Knudsen's notation 



dp 

 2jiR (M + B) -f .^i^^ \ — 0. 

 dl 



dp 



— is here the pressure-gradient, A' the radius, M the tangential 



dl 



force per cm', e.xerted by the gas on the wall in consequence of 



1) In controlling the calculations it was found that small errors aiid uncertainties 

 remain about the data for calculating the gas contained in the capillary, which 

 can change the numbers for the uncorrected T by some thousandths of a degree. 

 The necessary corrections are inside the limits of the experimental errors. So we 

 have left them mixed up with the latter. As soon as we shall have an opportunity 

 to compare the present determinations with more accurate ones, we can perhaps 

 return to this point. (Added in the English translation). 



