PROFESSOR THOMSON ON THE ELECTRO-DYNAMIC QUALITIES OF METALS. 717 
we imagine a rectangular zigzag, from side to side of the bar, instead of the true 
rectilinear course of the current, the current would be from transversely stretched to 
longitudinally stretched through hot. Hence it is established by this experiment, that 
iron, under a simple longitudinal stress, has different thermo-electric qualities in 
different directions. 
Knowing, as we do, from the first experiment on copper, described above (§ 107), 
that iron is not the only metal thermo-electrically affected by stress, we may con- 
clude with much probability that, in general, metals subjected to stresses not equal 
in all directions will acquire the crystalline characteristic of having different quali- 
ties, as regards thermo-electricity, in different directions. 
118. The qualitative investigation of the thermo-electric effects of stress, unaccom- 
panied by permanent strain, that is, the elastic thermo-electric effects of stress, would 
be complete for iron if the thermo-electric effect of a uniform dilatation or condensa- 
tion in all directions had been ascertained. I hope before long to be able to carry 
into effect various plans I have formed with this object in view ; but in the mean time 
it would be the merest guessing to speculate as to the result. 
119. The establishment of the crystalline characteristic for the thermo-electric 
effects of stress not equal in all directions, would make it probable that any thermo- 
electric effects which a metal permanently strained by such a stress can retain after 
the stress is removed, must also possess the crystalline characteristic. That this is 
really the case I had in fact proved, before performing the decisive experiment, just 
described, regarding the nature of the elastic effect, which was only made a few weeks 
since. The following experiments on the thermo-electric effects of permanent strains 
in metals were all made more than a year ago. 
120. Well-annealed iron wire was rolled in a coil of about twenty turns on a flat 
bar of iron |^-inch thick and 2 inches broad. The bar was laid on an anvil, with 
little pieces of thicker wire laid upon it to support the iron core and prevent the 
lower parts of the coil from being pressed. The upper parts of the coil lying on the 
upper flat side of the core were hammered till they were all very much flattened. 
The coil was then a little loosened and drawn off the bar of iron, and a similar wooden 
core was pushed into it. The ends of the iron wire were arranged, with the usual 
precautions (§ 92), in connexion with the electrodes of a galvanometer. A piece of 
hot glass (not above the boiling-point of water) was laid along one edge of the coil, 
so as to heat the iron wire at one set of the points separating hammered from 
unhammered portions. The galvanometer showed by a great deflection of its needle 
a current through the iron coil from hammered to unhammered through hot. When 
the heater was applied at the other edge of the flat coil, the deflection soon became 
reversed ; still, and always in subsequent repetitions, indicating a current from the 
strained to the soft metal through the hot junctions. 
121. The coil was next replaced on its iron core, heated to redness in the fire, and 
cooled slowly. It was then insulated by slipping in paper between it and the iron 
5 B 2 
