Metallic Vapours in an Exhausted Space. 105 
which may assist the propagation along the tube.~ This 
explanation is, however, not sufficient tc account for the 
peculiar action of the carbon filament. The following expe- 
riment is instructive in this respect. 
Experiment 9.—The arrangement is the same as in fig. 5, 
with the exception that the carbon filament is surrounded by 
a narrow glass tube open at both ends. When the arc starts 
up it first fills the inside of the narrow tube, gliding along 
the filament, although the mercury arc has in general the 
tendency to avoid narrow tubes, in view of the high resistance 
of the path. After a while the are, of course, spreads out 
and fills the whole section of the tube. 
The action of the carbon filament cannot be ascribed to the 
well-known ionizing-power of incandescent filaments, since 
the whole process of starting in a well-exhausted tube is 
nearly instantaneous, and the filament has not the time to get 
hot at all. After the are is started the filament does not 
play any roleat all. In consequence of its high resistance in 
comparison with that of the are, the current fowing through 
it is insignificant. The carbon filament is about the only con- 
ductor that can be used for this purpose, other filaments, 
such as iron, platinum, &c., being too good conductors and 
melting off the moment the arc strikes their lower end. 
§ 3. 
Properties of the Mercury Arc and of the two Electrodes. 
In so far as the discharge through mercury vapours is 
characterized by a relatively low voltage across the terminals, 
by a high current, and by a higher temperature of the anode 
in comparison with the cathode, this discharge must be 
classified in the same group with the arc between two pieces 
of carbon, or between twe metals in air. In consequence, 
however, of the high vacuum and the volatility of mercury, 
the arc can be made of any desired length, and offers the 
best opportunity for an exhaustive study of the subject. The 
are, as such, has been investigated in some respects by Arons; 
and, so far as his results go, I can in most cases confirm 
them. 
In common with all the ares, the voltage across the mercury 
are varies only little with the current. If, for instance, the 
current increases four times, the voltage will vary by only 
5 to 6 per cent. In the first approximation, therefore, the 
resistance of the arc can be considered as inversely proportional 
to the current (Arons). In contradistinction to the ordinary 
carbon arc, however, the voltage varies in the same sense as 
the current. This difference is probably explained by the 
