Heat across Layers of Gas. 425 



extending the investigation to the second branch of the sub- 

 ject, viz. the transfer of heat which accompanies those mecha- 

 nical actions. 



2. When gas is in contact with a body A at a different 

 temperature from itself, it is a familiar fact that convection 

 currents rapidly set in. The first step of the process is the 

 almost instantaneous formation of that layer which I have 

 called Crookes's layer — a layer of the gas of varying density 

 and temperature, being on one side at the temperature of the 

 body A, and on the other side at the temperature of the 

 surrounding gas. It is because this layer has a different 

 density from the rest of the gas, and because of the attraction 

 of the earth, that those streams set in w^hich are called con- 

 vection currents ; and accordingly, if the experiment could be 

 made at a station where there is no gravity, these convection 

 currents would not arise, although the Crookes's layer would 

 then also be fully developed. It will be convenient to inquire 

 first what will occur under these simplified conditions^ and 

 afterwards to take into consideration whether any modification 

 has to be made to allow for the effect of the neighbouring 

 earth. To give to the problem definiteness and the utmost 

 simplicity, I will suppose that a body A at temperature 6i 

 presents a large flat surface to an atmosphere of gas which is 

 at a lower temperature 62^ ^^^ exposed everywhere to a con- 

 stant pressure, but which is uninfluenced by gravity. Let us 

 further regard this gas as a perfect non-conductor of heat. 



3. If the excess of temperature is supposed to be suddenly 

 imparted to A, there will be a brief interval of adjustment 

 within the gas, after which the condition of the gas will settle 

 down into the state in which the Crookes's layer w^ill have 

 been fully formed. The Crookes's layer in this case will ob- 

 viously consist of a flat stratum of the gas in contact with the 

 hot surface of A ; and within this stratum the temperature 

 will gradually decrease from within outwards, from 6^, the 

 temperature of A, down to O2, the temperature of the sur- 

 rounding gas. This gradual falling-off of the temperature 

 implies a corresponding gradual augmentation of the density, 

 since we have supposed the gas to be everywhere subjected 

 to the same pressure. If the gas could * admit of the forma- 

 tion of a comiolete Crookes's layer, then we know, from the 

 familiar experiments which show gases to be bad conductors of 

 heat, that after the brief interval of adjustment a permanent 

 state would ensue, in which there would be no further change 

 of density, or motion of heat except by radiation. Accord- 

 ingly, if an isothermal surface be now drawn within the layer 



* See page 427, the last paragraph of section 4, 



