Heat across Layers of Gas. 



4^9 



in the upper line of fig. 9, to represent the observed excess of 

 the escape of heat over what was accounted for by radiation 



: nf 'h^^ 



mw'^%0^ 



mmm 



^m^r^l 



fel^-^"^: ^< 



convection 



we may draw the dotted parabola crossing it somewhere to 

 the right of A, and conclude that the convection, if it could 

 be separated from penetration, would be represented by the 

 ordinates of such a curve, and that the shaded interval between 

 the two curves is due to heat having leaked away by pene- 

 tration, across even so much as 12 centims. of hydrogen. 

 Although we cannot assign the exact position of the dotted 

 line until observations shall have been made in larger receivers, 

 we can already see the general shape of the shaded space 

 which is at present the object of our search. 



17. In a mixture of air and hydrogen within the cylinder, 

 De la Provostaye and Desains found that at a tension of 60 

 millims. the rate of cooling was ^^ much less " than if the hy- 

 drogen alone had been present. From this, and from the 

 observations plotted down in fig. 8, it follows that the Crookes's 

 layer in such a mixture is narrower than if all the molecules 

 present had been hydrogen. This accords with the theoretical 

 consideration that the members of both the processions in the 

 hydrogen will be more diverted from their course, and will 

 therefore transmit forward less of their own momentum and 

 vis viva, if some of the molecules they encounter are the hea- 

 vier molecules of other gases. 



18. Several phenomena observed by Grove, Tyndall, Mag- 



