and Heat -conduction in rarefied Gases, 55 



partition is equal to the mean velocity of translation of the mole- 

 cules passing through it in both directions. But one half of 

 these come from the partition, and have consequently a mean 

 velocity of translation <w' ; the other half come from the inte- 

 rior of the gas, and therefore have a mean velocity of translation 

 <w' . Consequently u Q <u' . 



A closer investigation, to be given in detail in another place, 

 has led to the following results. 



1. The sliding-coefficient for a gas and a solid partition has 

 sensibly a determined value dependent on the nature of the gas, 

 so long as the latter is present in layers thicker than fourteen 

 times the mean length of path ; and it is inversely proportional 

 to the pressure. 



2. The absolute value of the sliding-coefficient is obtained by 

 admitting that the gas-molecules are reflected from the partition 



with its velocity of translation, to l"4 r , — consequently for air, 

 for which at 760 millims. pressure /= 0-000083 millim. accord- 

 ing to Stefan, toO'000058 , p being the pressure in millims. 



of mercury. 



Prom our experiments it comes to about twice as much — 



namely, in round numbers, to 0*0001 millim. From this 



we may conclude that on the collision of the molecules with the 

 partition the velocities of translation of the two are not perfectly 

 balanced. By means of this value, it is easy to satisfy ourselves 

 that in Maxwell and Meyer's experiments the effect of the sliding 

 cannot have become sensible. 



Our experiments on friction were conducted according to 

 Maxwell's method ; for we measured the logarithmic decrement 

 of the torsional vibrations executed by a disk between two fixed 

 disks sufficiently near to it. As we made the weight of the glass 

 disk (diameter 159 millims., weight 61*9 grammes) so small 

 that it was borne by two very thin silver wires (diam. = 0'0630 

 millim.) in a bifilar suspension, we succeeded in making the dis- 

 turbing damping moments so slight that they might be neglected: 

 with the smallest decrements measured they only amounted to 

 a little over 1 per cent., and with nearly all the measurements 

 employed for calculations much less than 1 per cent., of the total 

 value observed. We thus obtained a substantial simplification 

 of the apparatus, one vibrating disk being sufficient. We will 

 here put, first, some measurements effected by means of our ap- 

 paratus, at pressures of 750 and 380 millims. 



