ir 
o> 
ON MOLECULAR PHENOMENA IN MAGNETISED IRON. 153 
its length, and the source of heat, such as a Bunsen flame, be moved 
along, an electric current is set up in the direction in which the flame 
travels. By means of clockwork the flame can be caused to move con- 
tinuously, and hence a continuous circuit is thereby obtained. There 
are, however, no signs of E.M.F. in the circuit until the recalescent 
point is passed; then reglow takes place behind the moving flame and the 
cooling effect in front. This thermal difference is, we believe, the cause 
of the resultant E.M.F., for it ceases when the flame ceases to move, 
and is absent in those metals where recalescence does not occur. 
VY. The thermo-electric position of iron undergoes a sudden change at 
the critical temperature. This was first noticed by one of us in 1875; 
twisting a platinum wire round the iron or steel wire under experiment, 
and connecting the free end of the platinum and one end of the steel wire 
to a galvanometer, a thermo-electric current was of course observed on 
heating the wire, but directly the jerk occurred in heating a sudden move- 
ment of the galvanometer needle simultaneously occurred, and similarly 
in cooling the thermo-current changed along with the anomalous expan- 
sion, and a moment after the iron regained its magnetic susceptibility. 
Hot iron is thermo-electrically negative to cold iron, but at the critical 
point a large increase in the H.M.F. is suddenly developed. Mr. H. 
Tomlinson ! has shown that iron at a bright red heat in contact with iron 
at the temperature of the air develops an H.M.F. of about one-twentieth 
of a volt, or upwards of twice that between a bismuth and antimony 
couple with a temperature difference of 100° C. between their junctions. 
Cumming was the first to notice long ago that the thermo-electric 
properties of iron changed at a red heat, but to Professor Tait’s classical 
papers on thermo-electricity we owe the first exact investigation of the 
changes that heat produces in the thermo-electric properties of iron. In 
his Rede lecture, delivered on May 23, 1873, Professor Tait remarks that 
when various pairs of metals were tried up to a red heat the thermo- 
electric diagram representing the relation of E.M.F. and temperature. 
always exhibited an anomaly when iron was one of the metals; at some 
temperature near a low red heat a change occurred, the ‘ Thomson effect’ 
being negative in iron at ordinary temperatures, became positive at a red 
heat, and remained so until a much higher temperature was reached, 
when another change of sign appeared to be indicated. ‘ Iron,’ Professor 
Tait remarks, ‘ becomes as it were a different metal on being raised above 
a red heat ; this may have some connection with the ferricam and fer- 
rosum of the chemists, with the change of magnetic properties and of 
electric resistance at high temperatures.’ ? 
VI. The electric resistance of iron at this temperature also changes ; 
Smith, Knott, Macfarlane, and more recently Hopkinson’ and Le 
Chatelier,® have published investigations on this point. Hopkinson finds 
a change in the temperature coefficient of the iron wire he used at 855° C., 
and of hard steel wire at a somewhat lower temperature. These tempera- 
to occur in other unannealed wires. This is an effect due to annealing by the 
flame, and disappears immediately, whereas the effect in steel is persistent. 
Proc. Phys. Society, vol. ix. p. 105 (Nov. 1887). See also on this point a paper by 
one of us (Newall, Camb. Phil. Soc., Jan. 1888.) 
? Nature, June 12, 1873. Trans. R.S.E., Dec. 1873. 
* Proc. Royal Society of Edin., Feb. 1875. 
* Phil. Trans. Roy. Society of London, May 1889. 
* Comptes Rendus, Feb. 10, 1890. 
