96 
PEOEBSSOE TYNDALL ON THE ABSOEPTION AND 
it is true, have supposed the upper atmospheric regions to be colder than space, the 
depression of temperature being due to the radiation of the aerial particles, just as a 
grass-blade is lowered, by its radiation, below the air which smTOunds it. This notion 
must, I think, be abandoned ; for, as far as experiment goes, it leads us to conclude 
that air, and particularly air in the higher atmospheric regions, behaves as a vacuum 
both as regards radiation and absorption. 
§15. 
In his paper on the conduction of heat by gases. Professor IVIagnus examines the 
question of convection, and has adduced some striking experiments to show that the 
cooling of an incandescent wire in hydrogen is not due to the convection of the gas. 
He finds that when the wire is enclosed in a narrow tube, with only a thin film of the 
gas surrounding it, and where therefore currents, in the ordinary sense, are hard to be 
conceived of, the gas still exercises its cooling power. It had often occurred to me to 
make this experiment; and when I first heard of its successful performance by Professor 
Magnus I adopted his conclusion, that the cooling was due to conduction. 
Peflection, however, caused me to change my opinion. Suppose the wire to be 
stretched along the axis of a wide cylinder containing hydrogen, we should have con- 
vection, m the ordinary sense, on heating the wire. Where does the heat thus dispersed 
ultimately go 1 It is manifestly given up to the sides of the cylinder. The transfer by 
convection is a transfer ultimately to the sides of the cylinder, and if we narrow our 
cylinder we simply hasten the transfer. The process of narrowing may continue till a 
tube like that used by Professor Magnus is the result ; the convection between centre 
and sides vrill still continue, and produce the same cooling effect as before. Whether 
we assume conduction or convection, the tube surrounding the wire must be supposed 
to possess sufficient conducting power to carry the heat off, otherwise it would become 
incandescent itself by the accumulation of the heat. 
The reasoning of Professor Magnus in connexion with this subject is of extreme inge- 
nuity. He contends that there is no reason why stronger currents should establish 
themselves in hydrogen than in other gases. Currents are due to differences of density 
produced by the expansion of a portion of the gas by heat. Now hydrogen actually 
expands less than other gases, and hence the differential action on which the cm’rents 
depend is less in this gas than in the others. Professor Magnus alludes to the friction 
of the particles against each other, but considers this ineffective. 
This reasoning leads us to the threshold of a question which might form the subject- 
of a long and profitable investigation. For a given difference of density, is not the 
mobility of hydrogen greater than that of the other gases'? The experiments above 
recorded, where different gases were brought into direct contact with the som-ce of heat, 
seem to answer this question in the affirmative. I have had no time to pursue the 
question regarding hydrogen; but I have made a few experiments, which show the 
influence of density on the mobility of a gas in a very striking manner. 
