106 Mr. G, J. Stoney on Polarization Sires* in Gases. 



of the layer, ami Sr the average difference of the velocities 

 forwards ami backwards at that station) is so rapidly on the 

 increase as quite to compensate in KprSv (the expression for 

 the now of heat, p being the density at the station, and K a 

 constant) for the diminishing density. During the second 

 period, i. e. when the molecules have become fewer than N, 

 the polarization is still on the increase, but not so rapidly 

 as before, and at the same time the flow of heat decreases to 

 zero ; for while p tends to zero as the exhaustion proceeds, 

 the polarization does not tend to infinity, but to a limit, viz. 



1 2 , where i\ and r 2 are the velocities corresponding to Tj 



and T 2 , the temperatures of the pistons. Now, when gas is po- 

 larized with this kind of polarization within a tube the sides 

 of which reflect the molecules, we can find limits between 

 which its thermal and mechanical properties must lie. 



9. Before proceeding to determine these limits, it will be 

 well to guard ourselves against making mistakes by passing 

 under review the orders of the several magnitudes with 

 which we are dealing in this inquiry. No accurate mea- 

 sures appear yet to have been made of the thickness of the 

 chinks of air or vapour on w r hich spheroidal drops rest. But 

 from approximate measures, some of w r hich were made by Mr. 

 Fitzgerald and some by myself, I think it may be inferred 

 that this thickness is somew T here about the thickness of a sheet 

 of paper (i. e. about a fourth-metre or the tenth of a milli- 

 metre) when a spheroidal drop of the density of water, at a 

 temperature of 10° C. and 5 or 6 millims. in diameter, floats 

 over a surface of liquid which is about 10° warmer. We 

 further know that in this case the Crookes's pressure, as it 

 supports the weight of this drop, must be about the two- 

 thousandth part of an atmosphere. These determinations are 

 very rude ; but they at all events tell us what kind of magni- 

 tude we are dealing with, and therefore suffice for our pre- 

 sent purpose. They show that we shall not be far wrong in 

 assuming definitively that the phenomenon presented by ex- 

 periment w r hich we have to explain is, that the stress across a 

 stratum of air will be 2 oV o P ar ^ °f the stress at right angles 

 to that direction, if this stratum occupies the space between 

 a heater and cooler at temperatures of 10° and 20° C, if, 

 moreover, this interval is a fourth-metre (a metre divided by 

 10 4 ), and if the atmosphere has free access to the stratum of 

 air at its edges. Let us now imagine a reflecting tube, such 

 as is described above, to be placed across this stratum. It 

 will therefore be a fourth-metre long ; and we may assign to it 



