Mr. G. J. Stoney on Polarization Stress in Gases. 403 



from these observations that after the brief interval of adjust- 

 ment is over the layer of vapour presses upwards and down- 

 wards more than it presses sideways ; for the pressure side- 

 ways must equal the pressure of the atmosphere so soon as the 

 adjustment is over, otherwise air would still be entering or 

 leaving the chink, whereas the pressure upwards must exceed 

 the pressure of the atmosphere by an amount able to support 

 the drop. It is my object to explain how this difference of 

 pressures, this Crookes's pressure as it has been called, comes 

 into existence. 



2. The thermal conditions of the problem are easily traced, 

 but need not detain us here. It is enough to state that 

 they show the metal dish and drop to be at different tempera- 

 tures, so that they are a heater and cooler on either side of 

 the layer of vapour. Experiment further shows that the 

 heater and cooler may be either one a liquid and one a solid, 

 as in the case already considered, or both liquids, or both 

 solid, and that the intervening layer may be either vapour or 

 permanent gas. This last important fact has been established 

 by Mr. Richard Moss in an admirable series of experiments 

 lately made by him to test the theory of the present communi- 

 cation (see ' Scientific Proceedings of the Royal Dublin So- 

 ciety,' vol. i. p. 89). It is also found to be immaterial whether 

 the heater or cooler is uppermost, or whether they face one 

 another sideways. 



Other facts of importance have been elicited by the experi- 

 ments at low tensions, of which the most significant are: — that 

 when the heater and cooler are maintained at given tempera- 

 tures the Crookes's stress between them may be increased 

 either by bringing the heater and cooler closer together, or by 

 attenuating the gas until a certain point is reached which 

 varies from one gas to another ; and that when that point is 

 passed, the force decreases and apparently without limit. 



3. We may express these facts in a very convenient form for 

 our present purpose if the heater and cooler are extensive flat 

 parallel surfaces at fixed temperatures. Conceive two exactly 

 similar patches on the heater and cooler directly opposite to 

 each other, and each occupying a unit of surface, and con- 

 sider that portion of space which lies between these. Then 

 the observations show that there is one definite quantity of 

 the gas to be left in the volume so marked out, if we wish to 

 produce the strongest Crookes's stress. And, further, by com- 

 paring Mr. Crookes's experiments on the mechanical action 

 with those of De la Provostaye and Uesains on the flow of 

 heat, we learn another important fact, viz. that the maximum 

 stress occurs when the quantity of gas is too little to admit of 



2D2 



