46 The N.Z. Journal of Science and Technology. [Jan. 
state and leave the iron shell behind, so solid iron bars at a red heat 
can absorb carbon into its structure to considerable depths, as in case- 
hardening. 
The ideal method of melting such a substance as iron sponge is the 
electrical, where a great temperature can be produced in a confined space 
under conditions which can easily be regulated so as to be either reducing, 
oxidizing, or neutral at will. To put this method to the test the furnace 
shown in fig. 3 was built. 
The reduction stage was carried out in practically the same manner 
as in the second furnace ; then by means of a sliding door the charge was 
allowed to fall into a plumbago crucible, which was then removed to one 
side and the charge melted down with an electric arc. 
The electrical supply was from a generator of 100 amps, at 100 volts, 
driven by a Westinghouse gas-engine with throttle-valve governor. Some 
experience was needed to manipulate the melting-arc, in the gaining of 
Fig. 3.—Final experimental furnace. 
which the governing of the engine was tested to the limit, but it behaved 
perfectly. The fluctuations in the current taken off were violent, running 
frequently from no-load to overload, and it was found necessary to put 
in a resistance of iron wire. With a little practice, however, it was found 
possible to keep the arc burning, with a good fat flame, for long periods 
with the resistance cut right out. The secret of the long, steady burning 
arc was found to lie in the fumes given off by the fluxing-materials, which 
formed a conducting-medium. 
A typical experiment was as follows : (1) Heat up whole reduction 
furnace with powdered coal; (2) charge hopper with 8 lb. ore and 8 lb. 
coal, and run through in fifteen minutes ; (3) run charge through into 
crucible ; (4) start arc, and maintain for one hour ; (5) add flux composed 
of 8 oz. of lime and 8 oz. of fluor-spar, and continue the arc for another 
hour and a half. Result: 4 lb. of metal and 3 lb. 3 oz. of slag. 
Except in one experiment (when Parapara limonite-ore was used) the 
resulting metal was always inclined to be white, due, I think, to the low 
silicon-content, as the conditions, of the furnace would not favour the 
reduction of silica. 
