75 
1912-13.] Magnetic Induction in Ferric Oxide. 
form to that obtained for the oxide when heated in hydrogen. But the 
curves differ in one particular. The magnetic property appeared sooner, 
and at a lower temperature, in those bars of ferric oxide which were heated 
Doubl e 
DefUcho r?. 
30 
n\rn. 
/Wer<a<?€ Curve for 3 bars of hroujq 
Ftrrlc Quid? Powder Cc fifier Tubz » 
Hoars. 
in air than in the bars heated in hydrogen. This holds both for the curves 
obtained at constant temperature and for those taken during continuous 
rise of temperature (compare Curve 4 with Curve 8, also Curve 1 with 
Curve 7). This result was entirely unexpected. 
3. Heating in Gas Furnace. 
The specimen was placed in a Fletcher gas furnace about one foot long, 
provided with ten Bunsen jets over its whole length. The bar did not 
come into contact with the flame, but was protected from it by means of 
a porcelain or copper tube. A thermo-couple with its junction at the 
middle of the bar was read every minute after the temperature had risen 
to 850°. When in the furnace, the bar lay east and west. After 10 
minutes’ heating at 850°, the gas was turned out and the bar cooled down 
to air temperature. 
Examination in the Magnetometer after Heating in Gas Furnace. 
(See Curves 5 and 6.) 
Some time after the surface of the specimen had become cold, it was 
removed to the magnetometer, and the deflection was noted which was 
caused by the bar on reversal of a given magnetic field. The above pro- 
cedure of heating in the gas furnace for 10 minutes, followed by exami- 
nation in the magnetometer after cooling, was repeated many times with 
each specimen, and the results were plotted in the following way : — The 
abscissa is divided into intervals which measure the 10-minute periods of 
heating. At the end of each 10-minute of abscissa, the height of the 
ordinate measures the deflection given by the bar after it had cooled down 
and was put into the magnetising coil. The curves given are for a bar of 
