of Edinburgh , Session 1872-73. 33 
the late Dr Mattkiessen. I reserve details until I can obtain the 
history and mode of preparation of the specimen examined, hut I 
may state now that, when formed into thermo-electric circuits with 
various alloys of Platinum and Iridium (Proceedings, 1871-2, p. 773) 
it gives results, as to the position of neutral points, not differing 
more from those given under the same circumstances by various 
iron wires of commerce, than the latter do among themselves. Thus 
it appears that, in the thermo-electric diagram, the line even for 
pure iron is sinuous ; and that the specific heat of electricity 
in it changes sign somewhere about a low red heat. 
3. Note on the Bate of Decrease of Electric Conductivity 
with Increase of Temperature. By D. H. Marshall, M. A., 
Assistant to the Professor of Natural Philosophy. Com- 
municated by Professor Tait. 
These experiments were undertaken in order to determine how 
closely the hypothesis “ that the electric resistance in a pure metal 
is directly as its absolute temperature ” holds for various metals 
at two easily ascertained temperatures, — that of the air in the room, 
and the boiling point of water. The apparatus used was a Wheat- 
stone’s bridge ; one coil of wire kept in a vessel of water at the 
temperature of the air in the room being put against another, which 
could be heated up to 100° C. The experiments showed that the 
rate of increase of resistance with temperature was different for 
hard and soft specimens of the same metal, being always less in 
the hard. This was further proved by additional experiments, 
which showed that sudden cooling always diminished the rate of 
increase of resistance, whereas if the metal were allowed to cool 
slowly after being boiled, the rate of increase of resistance was 
always sensibly increased. 
The first two columns of figures give the ratio of the resistances 
at the two temperatures ; the first and third give the ratios of the 
temperatures themselves in absolute scale ; the fourth is the differ- 
ence between the second and third, which will therefore show the 
amount and direction of deviation from the above hypothesis. 
When the number in the fourth column is +, the rate of increase 
