81 
1912-13.] Magnetic Induction in Ferric Oxide. 
How far this effect concerns the given curves may be judged from the 
following particulars : — 
Of the five bars which went to make Curve 1, two were moulded with 
alcohol and one was compressed, and the shortening in each was 14 per cent. 
In the other two bars, which consisted of oxide held in a copper tube, no 
shortening took place. Thus the final correction for Curve 1 amounts to 
about 9 per cent., or an increase of less than \ mm. Alcohol was also 
employed in forming the bars from which Curves 8 were obtained, but the 
temperature in this instance did not rise above 500° C., and no shortening 
was observed. In all other cases, viz. solid haematite bars and dry ferric 
oxide held in a copper tube, there was no shortening. 
From the foregoing it may be seen that this effect, although large in 
bars formed with alcohol, does not seriously affect the given curves. No 
correction has been applied for it. 
(c) Errors in Temperature Measurement . — The moving-coil temperature 
indicator used in conjunction with the thermo-couple had been standardised 
by means of a series of fixed points ranging from the ice-point up to the 
melting-point of potassium carbonate at 883° C. The scale divisions of the 
indicator had a value of about 7°, and were estimated to tenths by the eye. 
Temperatures could therefore be read to about 1° C., but we have not taken 
our readings to less than 5°, partly because of a variation of temperature 
from end to end of the furnace, and partly because, in work with the solid 
haematite bars, of the lag of temperature within the bar. The former error 
was removed in certain cases by the use of the copper tube. 
Temperatures above 883° C. were extrapolated. It has to be remarked 
that the temperatures in Curves 2 and 3, which rose to 1200° C., were 
measured by means of a platinum thermometer. 
Of the results submitted, those obtained by heating ferric oxide and 
haematite in air, although somewhat irregular, all agree on the point that 
magnetism appears with heating. This is contrary to the effect to be 
expected when ferric oxide is heated in an oxidising atmosphere. The less 
permanent nature of the effect in the case of the powdery ferric oxide as 
compared with that in the more solid haematite bars indicates that the 
effect takes place only at parts of the test specimen which are inaccessible 
to the air surrounding it. In the first case, the magnetism must be caused 
by an iron oxide ; and in the second, the curves for haematite, when 
compared with those for magnetite already published, indicate that 
ferrosoferric oxide is present. This has, however, not yet been verified by 
means of chemical analysis. 
The outstanding result of the heat cycles in hydrogen is the definiteness 
VOL. xxxm. 6 
