§&*•] VISCOSITY OF GASES. 301 
iquiry as far as the present section indicates. Inasmuch as I have 
roved that as R decreases from my largest to my smallest radius, the 
nine of F(6'% which holds for P— p=0, increases and finally merges 
lto (1 -f a 6")* independently of pressure, I have accomplished the main 
urposes of this section. 
Intimately connected with the present discussion is the occurrence of 
urface condensation of gases on platinum. If the law of inverse 
quares holds, then there will be no tendency of gases to condense in 
be capillary caual, however finitely small. This follows from the con- 
tant potential of a homogeneous elliptical shell (of which the capillary 
ube may in a special sense be said to be compounded) on any points 
iclosed within it. But in the case of the law of inverse squares or 
ny higher law, there may be condensation infinitely near the surfaces, 
ogether with the possibility that molecules condensed on the walls of 
he tube may again find their way into the canal. The occurrence of 
aol ecu] ar aggregates in the canal or transpiring current of gas in virtue 
f surface action of the platinum can not therefore be assumed to be nil. 
Kayser 1 finds that the height of the condensation atmosphere of am- 
nonia on glass inay exceed ,0002 cm . Compared with the radius of my 
pillary tubes (i£=.008 cm ), this is by no means small. Hence there is 
beason to infer that the increased molecular aggregation of the trans- 
>iring gas is not negligible. In conformity with the results of Kayser 
md of Chappuis 2 , O. Schumann 3 altogether rejects the transpiration 
nethod. I believe these strictures much too severe. Schumann's own 
)bservations are made at temperatures not exceeding 100°. Neither 
Tom these nor from any other earlier researches can the transpiration 
3ehavior of a gas between 400° and white heat be safely predicted. In 
cables 81 to 88 I have given the data in detail in order to show the ex- 
cellent accordance of the data among themselves, even if an hour or 
more of heating to redness intervenes. The differences are clearly 
errors of observation. If, therefore, condensation discrepancies are 
present, their time of variation must be almost instantaneous. 
Furthermore, the identity of law (l+« 0) 1 for both air and hydrogen 
is most easily interpreted as an immediate fact. Otherwise it will be 
necessary to suppose that air and hydrogen as regards condensation on 
platinum behave identically 5 or that possible differences of the law for 
air and hydrogen are just compensated by the condensation discrep- 
ancy. Both of these alternative hypotheses are exceedingly improb- 
able. Again, the law (1+a #)* applies for hydrogen at all temperatures 
between zero and white heat. It is therefore clear that in case of 
marked condensation the simple law in question would Shave to be re- 
placed by a much more complicated function. If, finally, the transpi- 
1 Kayser: Wied. Ann., vol. 14, 1881, p. 450. 
2 Chappuis: Wied. Ann., vol. 8, 1879, p. 1. Cf. J. W. Langley : Zeitsclir, f. Phys. 
Chem.,vol.2,1888,p.83. 
3Q, Schumann; Wied. Ann., vol. 33, 1884, pp. 381, 385. 
(955) 
