A Xote on Thermal Diffusion. 249 



temperature gradients at a point (x, y, z) are related to one 

 another by three (x, y, z) equations of the form 



^(4-) = -^(-^)=^riogT).. . (a) 



Consequently it follows that, over a distance within which k t 

 may be regarded as constant, v 1 l(vi+v 2 ) varies as k t \og T; since 

 k t is positive it is clear from this that the heavier gas should 

 be relatively denser than the light, towards the cooler end of 

 the vessel. The demonstration of this effect seems to afford 

 the best method of verifying the theory, and our present 

 object is io describe briefly some experiments undertaken 

 with this end in view. 



The experiments were made by the second author of this 

 Note, and definitely establish the existence of the pheno- 

 menon of thermal diffusion. Pressure of other duties un- 

 fortunately rendered it impossible to perfect the apparatus 

 so that it should be capable of yielding results of close 

 numerical accuracy; the quantitative effect on the pheno- 

 menon, due to the various modifying factors mentioned 

 above, was consequently not determined : thus, while the 

 predicted sign and order of magnitude of the effect were both 

 confirmed, the verification of the theory is to be regarded as 

 mainly qualitative. Further experiments are now in progress 

 at South Kensington, in which optical methods are being- 

 used for determining the composition of the gases before and 

 after the action of thermal diffusion ; by this means it is 

 hoped to obtain observations of sufficient accuracy to enable 

 the somewhat complex theory to be checked in all its more 

 important features. The results of this second series of 

 experiments will be published in due course. 



The experiments of this paper fall into three groups. Those 

 in the first group were made with a simple apparatus consisting 

 of two glass bulbs of about 100 c.c. capacity, connected by a 

 short tube provided with a stopcock of 5 mm. bore. This was 

 sufficient to ensure equality of pressure in the two bulbs when 

 opened, and to allow the diffusion process to proceed with fair 

 rapidity. Due provision was, of course, made for the initial 

 evacuation of the apparatus, the introduction of the gases to 

 be experimented on, and for drawing these off to be analysed. 

 The hot bulb was heated up to a temperature of about 230° C. 

 by means of a small sand-bath. The gases used with this 

 apparatus were carbon dioxide and hydrogen ; they were 

 chosen because of the ease with which they can be estimated, 

 and because of their large mass-ratio (22). 



