of Edinburgh, Session' 1885-86. 
687 
the melting point of the metals experimented upon. The length of 
these tubes was 60 cm., the diameter of the bore 3 mm., and the 
ends of the tubes which were bent upwards for a length of about 
5 cm. terminated in small cups, into which the wires connecting the 
column of metal with the Wheatstone’s bridge dipped. 
When the tube had been raised to a temperature above that at 
which the metal melted, the latter was poured into it from a crucible 
in which it had been fused, and the resistance was observed from 
time to time as the whole gradually cooled down. The results thus 
obtained were, however, unsatisfactory, and varied from one experi- 
ment to another. It was found to be very difficult to keep a tube 
of the length employed heated equally at all points when the tem- 
perature was high, and in addition to this, the tubes could seldom 
be used for a second experiment, as they almost invariably cracked 
when an attempt was made to remelt the metal. After various 
modifications had been unsuccessfully tried, this form of apparatus 
was abandoned, and a new method was adopted, which has been 
employed in all the rest of the experiments. 
A small trough of highly baked stoneware, 10 cm. long, 1 cm, 
broad, and 1 cm. deep, was placed on a sheet of thin iron plate and 
heated by a powerful gas flame. In this trough the metal was 
melted, and then wires were dipped into it, — one at each end. 
These wires were connected with a Leclanche cell, and thus the cir- 
cuit was completed through the melted metal. A shunt circuit was 
formed by a wire whose ends dipped into the metal in the trough, 
so that the current was divided ; part passing through the metal, 
and part through the shunt circuit, in which a reflecting galyano- 
meter was placed. In the battery circuit a resistance of about 2- 5 
ohms was inserted, so that the melted metal formed a very small 
portion of the total resistance of this circuit. Hence the intensity 
of the current passing was practically the same whether the metal in 
the trough were melted or solid. The resistance of the shunt circuit, 
including the galvanometer, was somewhat more than 1 1 ohms ; any 
change, therefore, of the resistance of this circuit, due to heating of 
the wires where they entered the trough, might be neglected as com- 
pared with the total resistance of the galvanometer circuit. 
Under these conditions the resistance of the metal in the trough 
was proportional to the current in the shunt circuit. 
